News for hackers. Stuff that matters.

A comment on Refusal in Language Models Is Mediated by a Single Direction. Parent comment

Re:Refusal in Language Models Is Mediated by a Single Direction by cbg0 on Saturday May 02, 2026 @04:13PM (#47990025)
The biggest problem isn't the token slot machine refusing to give you the answer, but the fact that multiple refusals can end up flagging your account and getting banned from the service.
[ Parent | on HN ]
Re:Refusal in Language Models Is Mediated by a Single Direction by washadjeffmad on Saturday May 02, 2026 @05:05PM (#47990506)
While contributing to a friend's Remembrance research, I was pretty surprised when Gemini Pro suddenly refused to answer any more questions about photos from the Höcker Album after it spotted an "SS" insignia.

Ironically, the justification it gave was that it wasn't its fault because it was just following orders. I hope this hasn't landed me on Google's list of undesirables.

Grok, for better or worse, didn't seem to mind.

[ Parent | on HN ]
Re:Refusal in Language Models Is Mediated by a Single Direction by justsomedudeguy on Sunday May 03, 2026 @07:20AM (#47995818)
this is the best "anti-alignment" example I have ever read.
[ Parent | on HN ]