logoalt Hacker News

verdvermyesterday at 11:56 PM1 replyview on HN

What happens if a model passes the government tests and then later someone fine tunes it to behave differently, without making their changes public?


Replies

sanxiynyesterday at 11:59 PM

Any reasonable safety testing should include finetuning and safety margin to account for others may do better finetuning.

show 1 reply