Knowledge Base Regulation Tracker Standards & Frameworks OSS Directory AI Incident Database Research Paper Library Case Studies Society Articles News Jobs About Charter Sign inBecome a Member
Home/AI Incident Database/Malicious pickle models on Hugging Face open reverse shells when loaded

Malicious pickle models on Hugging Face open reverse shells when loaded

Security researchers found around a hundred models on the public hub whose pickle-serialised weights executed code on load, including a PyTorch model that connected back to a remote host, demonstrating model files as a malware delivery vector.

Severity: MediumPickleReverse shellModel hubsafetensorsArtifact scanning

Contribute

Spot an error, or know a record we should add? Members can propose additions and submit corrections.