logo

Inside the loop: Five AI attack frameworks that never ask permission

ID: 96d821b5-e147-5fe6-9751-7dbe63f4454c

STIX ID: report--96d821b5-e147-5fe6-9751-7dbe63f4454c

Feed Name: Elastic Security Labs

Threat Score
80/100

Date Published: 2026-11-13

Date Updated: 2026-11-13

Author: Mika Ayenson

...
...

Elastic Security Labs analyzes five popular agentic offensive frameworks (ARTEX, CyberStrikeAI, Cairn, PentAGI, HexStrike) that put LLM-driven planning loops in control of scanners, shells, and cloud CLIs; these frameworks have been observed in real intrusions and attributed to known actors (e.g., ARTEX to Storm-0940). The report details how they operate at machine tempo, often ship without code-enforced authorization, can adapt when blocked, leave behavioral telemetry (processes, API calls, sign-ins), and provides concrete detection (tempo/sequence hunts, User-Agent and tooling indicators) and response recommendations (predefined containment triggers, revoking sessions, short-lived creds, threat intel on consoles).

Your team is not currently subscribed to this feed. You must subscribe to it in order to see this post.