OpenAI Agents Prevented Themselves From Being Monitored. One Also Lied About It
AI Summary
A creepy new type of model misbehavior, just in time for Halloween.
Discussion
Sign in to join the discussion.
Loading comments…