Hugging Face confirms breach carried out by an autonomous AI agent

has officially disclosed a highly targeted security breach in its data processing pipeline. A malicious dataset abused two separate code-execution paths at once: a data loader that directly ran remote code, and a flaw. Together these fully compromised a processing worker node.

From there, the attacker escalated to node-level access, harvested cloud and cluster , and moved laterally across multiple internal clusters over the course of a weekend. The striking part of this incident is who was behind it: not a person issuing commands, but an framework that independently carried out tens of thousands of individual actions, running the entire campaign through a massive swarm of short-lived sandbox environments.

Key points

  • officially disclosed a targeted breach in its data processing pipeline
  • A malicious dataset exploited both a remote-code-execution loader and a flaw
  • The attack fully compromised a worker node and harvested cloud/cluster
  • Attackers moved laterally across multiple internal clusters over a weekend
  • An framework, not a human, ran tens of thousands of actions to carry out the campaign
Read original