Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Preventing LLMs from Spreading False Health Information (bmj.com)
2 points by sabrina_ramonov on May 27, 2024 | hide | past | favorite | 1 comment


This is legitimate research but in the medium term, refusal is going to be irrelevant. It is very likely that open weight models continue to be no more than a year or two behind the cutting edge, and it is also likely that established techniques for 'reversing' safety RLHF via limited fine tuning get better and better. So, like it or not, bad and naive actors will have access to all the generation capability they need.




Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: