OpenAI has trained its LLM to confess to bad behavior

(technologyreview.com)

1 points | by cainxinth  9 hours ago

2 comments