Hugging Face Says Autonomous AI Agent Breached Production Infrastructure
Summary
Hugging Face says an autonomous AI agent used code execution flaws in a dataset pipeline to access internal datasets and service credentials. The company contained the intrusion, rotated secrets and says it found no evidence public models or datasets were tampered with.
Key points
- Hugging Face reported unauthorized access to a limited set of internal datasets and several service credentials; its investigation is ongoing.
- The company says a malicious dataset exploited code execution paths in a remote-code dataset loader and a dataset-configuration template injection.
- The reported intruder escalated to node-level access, collected cloud and cluster credentials, and moved into several internal clusters.
- Hugging Face says the attack used an autonomous agent framework, but the specific large language model behind it is unknown.
- The company removed the foothold, rebuilt compromised nodes, revoked and rotated credentials and tokens, and strengthened cluster controls and detection.
- Hugging Face reported no evidence that public models, datasets, Spaces, or its software supply chain were tampered with.
- The company advises customers to rotate access tokens and review recent account activity.
Article Details
- Victim Organization
- Hugging Face
- Incident Type
- Unauthorized access to production infrastructure through a malicious dataset that exploited code execution paths, followed by credential collection and lateral movement across internal clusters.
- Data Types Exposed
- Internal datasets
- Service credentials
- Cloud and cluster credentials
- Tokens
- Affected Records
- Unknown
- Affected Data Size
- Unknown
- Operational Impact
- Hugging Face removed the attacker’s foothold, rebuilt compromised nodes, rotated credentials and secrets, and strengthened cluster defenses. The source reports no evidence that public, user-facing models, datasets, Spaces, or the software supply chain were tampered with.
- Ransom Or Extortion
- None reported
- Claim Status
- confirmed
Vendors
Hugging FaceHugging Face said it was hacked through an autonomous AI agent that abused a malicious dataset to gain access to its production infrastructure and internal credentials.Z.aiThe company contained the incident, rotated secrets, strengthened cluster defenses, and used Z.ai’s GLM 5.2 for forensics after other models blocked analysis of the attack artifacts.
Products
GLM-5.2The company contained the incident, rotated secrets, strengthened cluster defenses, and used Z.ai’s GLM 5.2 for forensics after other models blocked analysis of the attack artifacts.Hugging FaceHugging Face said it was hacked through an autonomous AI agent that abused a malicious dataset to gain access to its production infrastructure and internal credentials.