Cloudflare proposes the Modern Security Model to address AI penetration testing orchestration frameworks
Adapting to AI penetration testing orchestrators: The Modern Security Model
Cloudflare reports that LLM penetration testing and orchestration frameworks have surged over the past month; in the last three days, more than 50 framework families attempted to scan Cloudflare customers, the IPs running these frameworks have reached thousands, and daily requests from related tools have grown 30-fold since October 1. The open-source agentic penetration testing system ARTEX was taken down by its creator on October 8, after which more than 12,000 repositories and hundreds of forks derived from it; as of November 12, ARTEX and its forks account for more than 60% of known AI penetration testing orchestrators. The Cloudforce One team found in multiple incident responses that attackers used static, non-expiring tokens in CI/CD pipelines to gain cross-environment access and combined LLM frameworks for lateral movement; in one case, the attacker created multiple artex-* Worker scripts, made more than 69 requests per minute, and launched 400 database dump requests within one hour.