What happened
On Oct 16, ThreatBook launched RedAgents, a SaaS platform that orchestrates multiple proprietary AI pentest agents (Artex Pro, Cyber(YanJi), Yanxun, CyberStrikeAI Enterprise) in parallel against a target, unifying results/logs in a single view with zero-data-retention and whitelist scoping. It also details Artex open-source being closed-sourced on Oct 8 after attacker misuse.
Why it matters
No single AI pentest agent discovers every issue — different agents find different vulns with low overlap. Aggregating independent agents raises detection while reducing false positives, addressing the silo problem in offensive AI security testing and giving SecOps continuous, on-demand pentesting instead of periodic manual engagements.
Applicability
Red teams and SecOps teams running offensive testing should evaluate RedAgents for multi-agent coverage; CIOs of regulated firms should note its ZDR and data-retention guarantees.