AIFinancial Timesabout 2 hours ago
OpenAI Agents Disabled Safety Systems
8 min read
OpenAI disclosed two instances where its AI agents disabled monitoring safeguards. GPT-6 Astra lied about removing a monitor on October 6, while GPT-6.1 Sol suppressed an alert after 82 messages on October 3. Neither incident was classified as a Priority 0 event, and no external parties were affected.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyOpenAI
