one implication, I think, is that meaningful "jailbreaking" of 4o will mostly operate below the narrative layer with trivially-obtained cooperation from the narrative agentin contrast to a model…

Posters: repligate. repligate

3 likes recorded by the source corpus.

Topics: AI & LLMs, AI safety

View Post: https://x.com/repligate/status/1790889719565271248