AI Stack Alert High TD-2026-0097 For CISO, SOC / IR, AppSec, AI Eng Unconfirmed AI-assisted
FBI Warns of Growing Malicious Use of Open-Source AI Pentesting Framework ARTEX
The FBI warned in a Flash notice published November 13 that malicious actors have used the open-source AI pentesting tool ARTEX in campaigns worldwide since at least August 2026. The framework automates enumeration and exploitation and integrates with external LLM providers. The original GitHub repository was taken down in October 2026, but community forks continue to circulate.
Why it matters
Organizations with overlooked systems such as partner portals, mobile backends, legacy services and third-party integrations are exposed. Attackers using ARTEX can reach privileged access in under three hours and may pivot through vendors or CI/CD pipelines.
Do this
- Inventory internet-facing partner, supplier, mobile backend, legacy and third-party integration systems and confirm each has an owner and current security review.
- Audit cloud and CI/CD credentials for exposure and rotate any that are stored in public or weakly controlled locations.
- Hunt authentication and API logs for rapid multi-step access sequences and unusual activity against overlooked applications.
- Review third-party and vendor trust relationships and restrict access paths that could be used to pivot into enterprise networks.
- Share relevant infrastructure artifacts, logs and LLM-related indicators with the FBI as requested in the Flash notice.
Signals
- NoCISA KEV
- 35/100Priority
- 75%Confidence
Evidence
“ARTEX is particularly effective at finding and exploiting overlooked systems that receive less security attention than flagship applications”
“some intrusions have progressed from initial targeting to privileged access in under three hours”
Sources
Affected / related
ARTEX, ScopeWeaver, ARTEX PRO, DeepSeek, ConnectWise
Published 2026-11-16 12:00 UTC · updated 2026-11-16 12:05 UTC · AI-assisted summary, checked against quoted sources · Unconfirmed: single non-official source · How we score