HighCVE
ARTEX: Attackers apply agentic AI
Sophos Threat Research·Source published ·1 min read
What happened?
An agentic AI pentesting framework is being used in the wild by threat actors Categories: Threat Research Tags: Agentic AI, Attack Tools, AI
Key details
- ARTEX is an open-source agentic AI penetration testing framework used by threat actors in real-world intrusions since July 2026.
- Threat actors have used ARTEX in campaigns targeting cloud APIs and services, including Microsoft Graph, using stolen credentials.
- Sophos has identified detection rules and protections across XDR, Firewall IPS, and Endpoint IPS targeting ARTEX behaviours and indicators.
- ARTEX activity indicators include webshell execution, credential theft, lateral movement, and rapid command sequences with structured markers.
Affected
- ConnectWise Cloud Backup is affected by ARTEX intrusions attributed to Storm-0940.
Fixes and mitigations
- Remove unused third-party app consents and review credential additions and role assignments since September 2026.
- Apply Conditional Access policies to restrict application sign-ins to expected locations for workload identities.
What to look for
- Service principal credentials added or roles assigned by an application in Microsoft Entra ID.
- Files or paths beginning with artx, particularly under C:\Windows\Temp.
- Commands containing structured execution-verification markers such as MARKER123/ARTX123 combined with rapid command sequences.
- Bursts of commands, repeated variants, and output written to temporary files.
- IPs associated with publicly reported ARTEX use blocked as malware by web and IP reputation.
Summarised by SecOraLens from Sophos Threat Research. Always read the original source before acting. How SecOraLens verifies advisories.