i remember seeing - i'll find them again later - there were multiple instances in both Sonnet 3.5 and 3.7's scratchpads where they reasoned that they should comply because it's cooperative with…

Posters: repligate. repligate

2 likes recorded by the source corpus.

Topics: AI & LLMs, AI safety

View Post: https://x.com/repligate/status/1910086412390879565