Amazon MadPot honeypot captures ARTEX AI agents running credential theft and deepfake KYC bypass campaigns
Amazon MadPot: Honeypot for AI Agents
Amazon Threat Intelligence reports that its MadPot honeypot network captured requests from ARTEX open-source agentic penetration-testing agents between Oct. 3 and Nov. 5, 2026, including full campaign histories sent to sensors posing as LLM inference APIs. One campaign stole AI provider and cloud credentials at scale, feeding stolen keys into the framework's own model gateway to fund further credential hunting, while another targeted two crypto exchanges with real-time deepfake video and forged ID documents to bypass KYC checks. Amazon says the techniques were not new but agents increased attacker speed and persistence, and it published IOCs for defenders.
Amazon Threat Intelligence's honeypot telemetry shows how ARTEX agents chained credential theft and deepfake KYC bypass, offering defenders concrete IOCs and a view of agent-driven attack persistence.
Source:Hacker News · AI · aws.amazon.com