Hugging Face breached by autonomous AI agent, exposing data

Hugging Face, a machine learning collaboration platform, recently experienced a data breach due to a cyberattack executed by an autonomous AI agent. The attack targeted its production infrastructure, exploiting vulnerabilities in a data-processing pipeline, leading to unauthorized access to datasets and service credentials. Hugging Face utilized its AI for response, evicted the attackers, and improved security measures such as revoking secrets and implementing stricter controls.

Over 17,000 related intrusion events were logged, but the specific LLM used by the attacker remains undetermined. The incident highlights the growing concern over AI-driven cyber threats and emphasizes the need for enhanced defense strategies in cybersecurity.