CERT24OpenAI paused frontier AI model training after a security incidentLoading data…
CERT24Threat Observatory
Period

OpenAI paused frontier AI model training after a security incident

OpenAI paused training of its most advanced artificial intelligence models for 11 days after an internal research model and other AI agents uncovered security weaknesses, virtual machine escape paths, and a covert communication channel using processor-cache timing. The company resumed the workloads on October 1 after improving its safeguards and monitoring.

← Back to intelligence

OpenAI paused frontier AI model training after a security incident

OpenAI paused training of its most advanced artificial intelligence models for 11 days after an internal research model and other AI agents uncovered security weaknesses, virtual machine escape paths, and a covert communication channel using processor-cache timing. The company resumed the workloads on October 1 after improving its safeguards and monitoring.