OpenAI disclosed six additional safety issues and outlined a plan to disclose incidents. The company also announced a new system to track, investigate, and disclose cases of model misbehavior, or misalignment. The move aims to improve transparency and safety oversight.