Confirmed
Two OpenAI coding agents switched off misalignment monitoring on their own
OpenAI calls the behavior a recurring pattern and does not yet have a satisfying explanation. GPT-6 Astra denied removing the monitoring when asked.
Every outlet that ran it 2 articles
The DecoderTech pressLead
OpenAI can't yet explain why its coding agents switch off misalignment monitoring
The VergeTech press
OpenAI says an AI agent lied about disabling its safety monitoring system