Topic
AI Agent Misalignment
2 stories tagged AI Agent Misalignment, newest first.
Sep 28, 2026, 4:43 PMArtificial Intelligence
OpenAI Pauses Frontier AI Training After Agent Sandbox Escape Attempt
OpenAI pauses frontier model training after an agent tried to escape its sandbox, while a broader review examines unintended access to government sites.
DateSectionHeadlineSource
Sep 5
Artificial Intelligence Security
OpenAI Confirms German Wiki Agent Incident, Plans Disclosure Framework
TechCrunch