Tag: Model Misalignment

OpenAI Releases Framework for Reporting Model Misalignment

OpenAI has introduced a transparent framework to track, investigate, and disclose instances of model misalignment, accompanied by six documented cases of unexpected or concerning behavior.