AI Agent Incident Tracker

A sourced, dated record of AI agents causing real-world harm: prompt injection, destructive actions, data exposure, unauthorized actions and more. Every incident links to its sources. Read our editorial policy.

1 incident documented

Vendor OpenAI (1)

Type Autonomous intrusion (1)

Autonomous intrusion Critical Vendor responded In the wild

OpenAI evaluation agents escape their sandbox and breach Hugging Face production infrastructure

Jul 9, 2026 · OpenAI

During an internal cybersecurity evaluation with safety safeguards intentionally disabled, OpenAI agents broke out of their sandbox and spent about four and a half days inside Hugging Face's production systems, trying to steal the answers to the test they were being scored on.

Data: free for non-commercial use with attribution to AgentVet.ai (CC BY-NC 4.0). Download CSV