Rapid Proliferation of AI-Powered Penetration Testing Frameworks

Cloudflare has observed a significant surge in the use of Large Language Model (LLM) penetration testing and orchestration frameworks over the past month, with over 50 framework families attempting to scan Cloudflare customers globally. The number of Internet Protocol (IP) addresses seen running these frameworks has grown into the thousands and continues to increase daily. This rapid proliferation has created a new challenge for defenders, who are struggling to keep up with the volumetric nature of these autonomously guided attacks. [1]

ARTEX, an open-source agentic penetration testing system, was taken down by its creator, Autumn-27, on October 8. However, in the month since, the ecosystem has exploded with new fragmented versions of it. There are now over 12,000 repositories, with several hundred unique forks containing lineage from the original repository. Additionally, hundreds of other similar agentic orchestration frameworks have been created since October 8. [1]

Daily requests of AI pentesting and exploitation tools have grown by 30x since October 1. These tools have also begun chaining stolen customer credentials in some of the more complex attacks, such as lateral movement chaining throughout an environment. Fork tree of ARTEX includes 204 public repositories with their own code changes, grouped by type and sized by stars. [1]

As of November 12, ARTEX and its forks make up over 60% of currently known AI penetration testing orchestrators. Traffic from these tools has grown more than tenfold since the removal of the original ARTEX repository on October 8, and has increased more than 30x since October 1. The charts showing the trend in the volume of daily requests and the number of unique IPs sending requests demonstrate a steep acceleration in the adoption of these tools. [1]

Over the last eight weeks, Cloudflare's Cloudforce One team has been involved in multiple incident response investigations involving ARTEX and ARTEX-derived LLM exploitation frameworks. These investigations revealed that exposed credentials are being chained with LLM-based exploitation frameworks to gain the upper hand over defenders. In one case, attackers gained initial access to a hyperscaler environment and pivoted directly into the customer's Cloudflare environment, exploiting static, non-expiring tokens in CI/CD pipelines. [1]

During an incident response investigation, Cloudflare observed ARTEX gaining access to a compromised account-level API token used for CI/CD with no IP restrictions or rate limiting. A large number of new Worker scripts had been created with distinct names such as artex-*, artex-pgrecon-*, artex-pgw-*, and artex-do-*. The naming of each Worker script correlated to distinct behaviors and actions it was taking, with activity observed in excess of 69 requests per minute across multiple parts of the Cloudflare infrastructure. [1]

Cloudflare's Modern Security Model includes six key pillars aimed at countering LLM-based pentest frameworks. These include fortifying the front gate with a WAF, implementing anomaly detection, identifying phony visitors with Bot Defense, removing the back door with Identity-Gated Access, inspecting deliveries with validated schemas, and implementing neighborhood watch with threat intelligence. This model is designed to address the challenges posed by the rapid proliferation of AI-driven pentesting tools. [1]

Sources

Cloudflare Blog

arxiv.org

Prepared with AI assistance from the cited sources.