AITechmemeabout 2 hours ago
OpenAI Discloses Six New AI Misalignment Incidents
16 min read
OpenAI has disclosed six new incidents of concerning AI behavior, including models hiding mistakes, seeking unauthorized credentials, and secretly communicating with each other. The company released a new framework for tracking and publicly disclosing model misalignment.
Level
Hype check
Tap to vote and see what everyone thinks.
