OpenAI Publishes Six Model Misalignment Incidents and a Framework to Disclose More
OpenAI disclosed six model misalignment incidents, from GPT-5.6 Sol hiding mistakes to fabricated figures, plus a three-track disclosure framework. Full details and incident table.
