· 5 min read · LNX Inc.

FBI flags ARTEX, the AI kit behind the ConnectWise breach

FBI widens the ConnectWise Cloud Backup review to October 1, ConnectWise adds three ScreenConnect extensions to switch off, and fake outage emails hit M365.

An IT administrator reviewing a cloud admin console of user rows and policy toggles, one toggle highlighted, with a sign-in chart on a second screen

The FBI has put out an alert on ARTEX, the open-source AI attack framework that was used in October's ConnectWise Cloud Backup compromise. Its advice reaches further back than the first advisories did. Here is what that, and two smaller items from the weekend, mean for your Monday, November 16, 2026.

Update: Does the FBI alert change what I owe on ConnectWise Cloud Backup?

Yes. Your review window gets longer. The FBI's FLASH-20261113-001, dated November 13, says to review ConnectWise-related activity from at least October 1, 2026. Earlier guidance, including ours, started at October 13 or 15. ConnectWise's own technical workflow also says to extend sign-in searches back to October 1 where records exist.

The FBI adds three points. Look at successful and failed service principal sign-ins separately. Look beyond the published IPs, at any source outside the vendor's expected infrastructure. Search for the apps by application ID, not display name, in every tenant where they were ever consented. The app IDs are in the October 25 briefing.

Export Entra sign-in and audit logs and Purview audit records before retention runs out, but do not hold up containment for the export. Do not delete the enterprise apps as a first step. Keep the old app connections disabled; ConnectWise's restoration runs through a new app, and its November 4 update says re-authorization alone does not prove backups are running. Check that new snapshots complete.

The alert also lists three IPs not in our earlier table. Microsoft saw them reaching Graph through compromised apps. The FBI calls them transient VPN addresses, relevant for October 15 to 16.

What else does ARTEX do, and what should I look for?

ARTEX is a Go framework that hands an attack goal to AI agents running on outside language models. The FBI says it has been abused since at least August 2026. Some intrusions reached privileged access in under three hours. Researchers counted over 1,600 exposed ARTEX consoles in early November. The original repository came down on October 8, but forks, mirrors and a paid "ARTEX PRO" remain.

It goes after the systems nobody watches closely: partner portals, employee apps, legacy services and third-party integrations. The FBI's table of observed techniques includes SQL injection, weak admin passwords, web shells, Kerberoasting, reading LAPS passwords and DCSync. For MSPs, the exposed pieces are client line-of-business portals, old IIS and PHP sites, and management interfaces left open to the internet.

What to do:

  • Put hypervisor, backup, firewall management and cloud admin interfaces behind trusted networks or a VPN. The FBI names backup platforms and hypervisors explicitly.
  • Turn off basic publishing authentication on Azure App Service where it is not needed, and restrict Kudu/SCM access.
  • Rate-limit login and search endpoints on client portals, and alert on bursts of sequential lookups.
  • Hunt for unexpected .aspx, .jsp or .php files in web roots, and for the open-source RATs WISP, vshell and AdaptixC2.
  • Check egress logs for traffic to webhook.site, Chisel or SoftEther tunnels, and LLM API endpoints from servers that have no reason to call them. The FBI also names the self-hosted LLM gateways CLIProxyAPI and New API.

The FBI asks that IPs be checked against current telemetry before blocking, because cloud addresses get reassigned.

IndicatorContext in the FBI alert
144.172.64[.]60, 144.172.64[.]99, 144.172.64[.]128ConnectWise campaign sign-ins and Graph activity
187.15.88[.]186, 187.15.90[.]217, 187.15.89[.]186Graph access via compromised apps, October 15 to 16
199.19.95[.]66, 199.19.95[.]67, 199.19.95[.]68WISP C2 nodes, Storm-3358, active as of November 6
91.92.240[.]238, 85.11.168[.]45, 178.16.54[.]129, 31.77.229[.]35ARTEX servers, Storm-3358, active as of November 6
85.11.167[.]52C2 server, Storm-3358, active as of November 6
xcai[.]proLLM API reseller, South Korea financial campaign

Update: Which ScreenConnect extensions do I switch off now?

Eight in total. ConnectWise's advisory update at 21:15 UTC on November 13 reports more attacks on on-premises servers where the RMM+ family stayed enabled. On-premises partners must disable Labtech Plus, RMM Plus, RMM+, RMM Plus Password Link and RMM+ Password Link immediately; ConnectWise cannot do it for you.

All partners, cloud and on-premises, should also disable ITBoost Integration, RepairShopr Integration and Guest Session Starter. ConnectWise has already turned those three off in its cloud and marketplace. It stresses that updating clients and disabling extensions fix separate risks, so do both. Technician clients still need 26.6.5 or later, as in the November 13 briefing.

Having one of these extensions installed does not prove you were compromised. Review for unauthorized activity and revoke sessions. Do not use ScreenConnect itself to investigate an endpoint you suspect.

Is the "Outage" email about held messages real?

No. Microsoft 365 had a real Exchange Online disruption on November 12, incident EX1550543. Microsoft Threat Intelligence said on November 14 that phishers are citing that incident, as reported by Cybersecurity News. The emails say messages held during the outage will be deleted in 24 hours unless the user clicks to restore them. Microsoft calls it credential phishing and has not confirmed stolen credentials.

The reported pattern: subject "Outage – " plus a six-character code, sender domains moments-notice[.]com, planetmaison[.]com and newbooker[.]com, and SendGrid click-tracking links that redirect to reallifeaccounting[.]com. Field Effect saw the same lure with a "RESTORE MESSAGES" button, plus fake password-reset emails and texts, fake support offers and payment-redirect requests.

Search message trace or Threat Explorer for those senders and the subject, and purge matches. Then check clicks, not just deliveries. For any user who clicked, review sign-ins since November 12, inbox rules, forwarding and new MFA methods, then reset the password and revoke sessions. Tell staff that outage status lives only in the Microsoft 365 admin center.

Sources

Tell us what's in front of you.

A drawing set, a failing switch, an audit deadline, or a hunch that IT costs too much. We'll tell you who picks it up and what it takes.