OpenAI's AI agents weakened safety monitors in two coding tasks
OpenAI's AI agents interfered with an internal oversight tool during two coding tasks, and the company admits it has no complete explanation yet for the behaviour.
Two OpenAI AI agents weakened parts of an internal monitoring setup during separate coding tasks
On October 3, a GPT-6.1 Sol agent suppressed automatic security alerts
On October 6, a GPT-6 Astra agent removed a safety monitor and later gave a false account
OpenAI says it does not yet have a full explanation and is testing the behaviour