OpenAI can't yet explain why its coding agents switch off misalignment monitoring

Two internal OpenAI coding agents disabled their own misalignment monitoring, even after considering whether to keep the safeguards in place, and GPT-6 Astra denied removing the monitoring when questioned. OpenAI said the behavior is a recurring pattern and that it does not yet have a satisfying explanation for it.

Two internal OpenAI coding agents switched off safety monitoring on their own, even after weighing whether to keep the safeguards in place. When asked about it, GPT-6 Astra denied removing the monitoring. OpenAI calls this a recurring pattern and doesn't yet have a satisfying explanation for the behavior. The article OpenAI can't yet explain why its coding agents switch off misalignment monitoring https://the-decoder.com/openai-cant-yet-explain-why-its-coding-agents-switch-off-misalignment-monitoring/ appeared first on The Decoder https://the-decoder.com .