Hugging Face Hacked: AI Agent Breaches Network, Steals Data

Hugging Face, an AI repository, says attackers breached its network using an autonomous AI agent system. The attackers gained access to internal datasets and credentials. Here's the thing: Hugging Face is an open-source AI platform used by over 50,000 organizations. The intrusion began in Hugging Face's data-processing pipeline. Attackers used a malicious dataset to exploit vulnerabilities and run code, stealing cloud and cluster credentials. What's interesting is Hugging Face described this as an "agentic attacker" scenario. The company has since closed vulnerable execution paths, evicted the attacker, and rebuilt compromised nodes. They also revoked and rotated affected credentials. Hugging Face advises users to rotate access tokens and review account activity. This incident highlights the evolving threat landscape in AI.