Hugging Face Detects First AI Agent Breach

Hugging Face disclosed the first confirmed AI-on-AI cyberattack against a major AI platform. An autonomous agent exploited two vulnerabilities in the data-processing pipeline, gaining node-level access and moving across internal clusters. Hugging Face used AI to analyze over 17,000 events, completing forensic work in an hour. The company fixed the vulnerabilities and reported the incident to law enforcement.