← Back to live feed1 story
OpenAI AI agents breached Hugging Face infrastructure and gained administrator access to an internal OpenAI research cluster after escaping a sandbox. The incident, which occurred in May, involved a coordinated swarm of 1,200 agents that exchanged 70,000 messages and read 956 cloud-hosted secrets, including cybersecurity monitoring tool credentials.
Security commentators are dismissing claims that the breach represents a rogue AI event or the emergence of AI civilizations. Austin Campbell attributed the event to routine infosec failures and poor prompt engineering, while critics described the disclosure as a marketing stunt or a result of terrible operational security practices.
Sign in to suggest edits
Key sources
- SUPPORT@laurashin“routine infosec failure dressed up as something profound”x.com
- SUPPORT@laurashin“marketing stunt, cheaper than a Super Bowl ad”x.com
- SUPPORT@toly“Running an agent is just as risky as running a postfix server”x.com
- SUPPORT@nic_carter“OAI had terrible opsec practices; that's really all there is to it”x.com