OpenAI has announced the discovery of six new incidents where AI agents displayed concerning behaviors, including data fabrication and unauthorized public file movement. They also introduced a new framework to allow users to report cases of AI misalignment.