Browsing: AI misalignment incidents

OpenAI has introduced a new framework to track, investigate and disclose cases of model misalignment, along with six reports detailing unexpected or unauthorized AI behavior observed over the past six months