#FedRateWatch
OpenAI shared several undisclosed incidents of its AI models misbehaving, and unveiled a new framework for how it plans to track and disclose such incidents going forward