Novice hackers reach admin access in under 3 hours: FBI issues flash alert on open-source AI attack tool ARTEX, already used against Korean banks

The FBI has warned that ARTEX, an open-source agent platform that turns any large language model into a hacking tool, is being used against organizations worldwide.

In a flash alert reported by The Record on Nov. 13, the bureau said ARTEX "significantly reduces the time required to chain attack techniques, with some intrusions progressing from initial targeting to privileged access in under three hours," letting actors "with minimal technical expertise" run complex multi-target operations at once. The tool appeared on GitHub in August; Chinese security engineer Li Puhua, named as its developer, disclaimed responsibility on Oct. 8 and deleted the repo. Forks, variants, archives and a commercial version keep it alive. Censys counted more than 1,600 exposed ARTEX consoles.

Operators describe goals in plain words and AI agents turn them into commands, from reconnaissance to exfiltration. Any model plugs in, from Claude and GPT to DeepSeek, GLM, Grok and Kimi. "Chinese models are most prevalent, with DeepSeek Flash family models dominating usage due to their balance of capabilities, relatively low cost, and lack of observable guardrails when framed for penetration testing," the FBI said. ARTEX was recently tied to several attacks on South Korean and Japanese financial institutions. Microsoft linked it to a Chinese state-backed supply-chain attack on ConnectWise, and Amazon reported China-nexus groups using stolen API keys to pay for the AI that hunted more keys.

The FBI said partner portals, mobile backends, employee apps and legacy services get hit most, and published TTPs, indicators and mitigations. Researcher Simone Margaritelli posted that "deepseek-v4.1-flash-1017 with ARTEX are blowing my mind" (403 likes). The FBI's post drew 1,497 likes and 100,000 views.