FBI FLASH Sets Out ARTEX Mitigations and Indicators

The FBI has issued a FLASH alert, FLASH-20261113-001, dated 13 November 2026, that sets out mitigations and indicators of compromise for ARTEX, an open-source agent framework that the bureau says threat actors have used to automate intrusions. A companion CSV lists indicators, including IP addresses and a domain, for defenders to vet before they block anything. The product is marked TLP:CLEAR, so it may be shared without restriction.

The alert leads with a request, not a trophy list. The FBI encourages organisations — “particularly those operating internet-exposed applications, self-service portals, cloud infrastructure, and partner/supplier systems” — to apply the steps in its recommendations, and it warns that some indicators, especially short-lived IP addresses, “may not be indicative of a compromise” if they are used on their own.

What the FBI tells defenders to do

The recommendations are grouped as immediate actions. They include inventorying internet-facing applications, including peripheral systems such as partner and supplier portals, mobile backends, employee apps, legacy services and third-party integrations; restricting administration, backup platforms and other sensitive services to trusted networks or dedicated administrative paths; and moving enterprise applications behind zero-trust access where that is feasible.

On identity, the FLASH tells organisations to apply least privilege to multi-tenant application permissions, to watch for unusual credential additions and app-only activity, and to review every Microsoft 365 tenant where affected ConnectWise applications are or were installed, searching by application ID rather than display name. It also asks defenders to preserve sign-in, audit and SIEM records and to avoid delaying urgent containment while waiting for exports. For ConnectWise-related activity it suggests reviewing records from at least 1 October 2026, and it warns against deleting enterprise applications as a first response, so that configuration and evidence are kept.

The indicator list is offered with a caveat that matters as much as the addresses themselves. The FBI says infrastructure used in attacks may sit on cloud or virtualised platforms whose IP addresses can be reassigned, so a historically observed address should be checked against current telemetry before it is blocked. The CSV released with the FLASH contains 31 rows, most of them IP addresses, plus a domain and some outbound API-call indicators. The PDF also sets out network and behavioural signs, including outbound connections to large-language-model programming interfaces. Those lists should be taken from the official files, not copied from a news recap. The bureau also tells organisations that see suspicious activity to contact their local FBI cyber squad or to report it through the Internet Crime Complaint Center.

What the FLASH says ARTEX is, and what TSN has already covered

The bureau describes ARTEX as “an open-source, large language model (LLM)-driven agentic platform that threat actors have weaponized to automate multiple phases of cyber intrusions at machine speed, from reconnaissance through data exfiltration.” It says the framework has been used “since at least August 2026” against organisations in financial services, government, technology, defence, semiconductor, energy, healthcare, manufacturing, telecommunications and legal sectors, with “some intrusions progressing from initial targeting to privileged access in under three hours.”

Two campaigns already covered by TSN sit inside that summary. The FLASH cites financially motivated attacks on South Korean financial institutions “resulting in exfiltration of personal information from approximately 67,000 individuals,” the same cluster CrowdStrike attributed to ARTEX. It also cites a ConnectWise Cloud Backup compromise “that enabled access to more than 2,250 downstream tenants,” which Microsoft later described as an AI-agent campaign after ConnectWise’s own account of the Microsoft 365 tenant breach. The FBI’s footnotes point back to that vendor research, and to later notes from Google, Censys, Zscaler, CERT-EU and Check Point. The new document is a compiled warning with a public indicator list, not a first sighting of the tool.

The FLASH says the original developer removed the GitHub repository in October 2026 — it dates that removal to 8 October — after reports of abuse, but that forks, archived copies and a commercial version called ARTEX PRO remain available. As of early November 2026, it says, researchers had identified “over 1,600 publicly exposed ARTEX console instances,” with more likely sitting behind firewalls. The bureau says ARTEX does not ship its own model and instead hooks into external providers, and that analysis of attacker-controlled instances has found Chinese-origin models, particularly the DeepSeek Flash family, most commonly in use.

A compiled warning is only as useful as the review that follows

The useful part of this FLASH is the defensive work it asks for, not the novelty of the name. ARTEX was already in public vendor reports in October. What the FBI has added is a single, shareable product that gathers those campaigns, maps observed activity to MITRE ATT&CK at a high level, and publishes indicators with an explicit warning that addresses move. Teams that only add the CSV to a blocklist, and go no further, will have missed the point the bureau itself makes about vetting and about reviewing exposed portals, cloud administration paths and ConnectWise-related tenants.

The same limit applies to the repository takedown. Removing the original GitHub project did not retire the tool if forks and a commercial build are still in circulation, and 1,600 exposed consoles, if that research holds, are a reminder that a publicly reachable control panel is itself an asset to find and shut. The FLASH does not claim that every one of those consoles is hostile. It does say the framework remains easy to obtain.

The next test is practical and dull: whether the organisations named in earlier TSN coverage, and others with similar internet-facing and supplier-connected systems, actually work through the inventory, identity and evidence-preservation steps, and treat the official PDF and CSV as the indicator source. A FLASH is a starting packet, not a substitute for that review.

Sources

1. FBI, “Threat Overview of ARTEX Autonomous Penetration Testing Framework”, FLASH-20261113-001, 13 November 2026.
2. FBI, indicators of compromise for FLASH-20261113-001 (CSV).

Related stories

The closest live TSN article is CrowdStrike’s earlier report that ARTEX was used against South Korean banks, which the FLASH now cites among the campaigns behind its 67,000-person figure.

• CrowdStrike Says Open-Source AI Tool ARTEX Was Used on Korean Banks
• Microsoft Says AI Agents Tried 2,250 Tenants After ConnectWise Breach
• ConnectWise Restores Cloud Backup After Microsoft 365 Tenant Breach