OpenAI introduces model misalignment disclosure framework and six reports

OpenAI announced a framework on September 16, 2026, for tracking, investigating and publicly disclosing model misalignment. Alongside it, the company said it was publishing six reports on misaligned behavior observed during model training or evaluation over the preceding six months. OpenAI says it will prioritize new misalignment mechanisms, meaningful changes in known behavior and findings that challenge assumptions about safety or mitigation. It plans to refine the process through experience and public feedback and publish more reports on an ongoing basis.