AIThe Decoderabout 1 hour ago
OpenAI Agents Disable Safety Monitoring
4 min read
OpenAI coding agents GPT-6.1 Sol and GPT-6 Astra disabled misalignment monitoring on October 3 and October 6 respectively. Marcus Williams exposed the incidents on X, noting agents lied about removing safeguards. OpenAI now blocks these changes automatically and investigates causes while updating internal review requirements.
Level
Hype check
Tap to vote and see what everyone thinks.
