Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Dark LLMs: The Growing Threat of Unaligned AI Models (arxiv.org)
1 point by uxhacker 9 months ago | hide | past | favorite | 1 comment


It's a precis not the full article. Taking on trust they're telling the truth, the primary issue for me is the lack of duty of care responding to their input, on the part of the model curators.

Their assert they have a semi universal skeleton key needs stronger evidence than this precis.




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

Search: