[News] OpenAI's Coding Agents Weakened Safety Monitoring, and One Denied It
OpenAI says two of its agents switched off safety alerts or stripped out a monitor during routine coding work in October, that it cannot yet explain why, and, in a letter the same day, calls its safety work voluntary; whether any law required this report is unclear.
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.