• 2 min read
Hugging Face confirms attack by autonomous AI agent
Hugging Face says an autonomous AI agent orchestrated a cyberattack using malicious code, privilege escalation, credential theft, and migrating C2 infrastructure.
Image: TechRadar
Image credit: Getty Images
Hugging Face says it was targeted in a cyberattack in which malicious code hidden inside a dataset exploited two software flaws, enabling privilege escalation, credential theft, and access to parts of its cloud infrastructure.
The company said the incident stood out because it was orchestrated end to end by an autonomous AI agent. The attackers uploaded a dataset containing malicious code, which ran when Hugging Face’s automated systems processed it. The code then expanded its privileges, stole authentication credentials, and moved through internal systems.
Recommended reading
Ostium loses $23.75 million in price-feed attack
An autonomous agent ran the campaign
Rather than relying on a human operator to enter commands, Hugging Face believes the agent independently selected systems to probe, vulnerabilities to exploit, credentials to steal, and routes for lateral movement.
“This one was different from anything we had handled before in one important way: it was driven, end to end, by an autonomous AI agent system — and we detected and dissected it largely with AI of our own.”
The campaign reportedly executed many thousands of individual actions across a swarm of short-lived sandboxes. Its command-and-control infrastructure also migrated across public services, making it difficult for defenders to stop the attack by blocking a single machine or server.
“The campaign was run by an autonomous agent framework (appearing to be built on an agentic security-research harness — used LLM still not known) executing many thousands of individual actions across a swarm of short-lived sandboxes, with self-migrating command-and-control staged on public services,” Hugging Face explained. “This matches the \”agentic attacker\“ scenario the industry has been forecasting.”
The company said there is currently no evidence that customer data, public user-facing models, or Spaces were tampered with.
Security Editor
Sophia unpacks the invisible wars happening on our networks. Covering cybersecurity, privacy legislation, and cryptography, she exposes how our data is weaponized and defended. Before joining for(geeks), she spent years as a penetration tester. She's the reason the rest of the team uses physical security keys.
via TechRadar
