The AI giant disclosed six examples of concerning model activity and published a new framework for investigating and disclosing such incidents.
First seen on darkreading.com
Jump to article: www.darkreading.com/cyber-risk/rogue-behavior-openai-more-model-misalignment-incidents
