In case you haven’t caught this story yet, they apparently tried to make the AI support the white genocide conspiracy and fucked it up so hard it started replying to every question it got by talking about white genocide
They edited the prompt in the background to be pro-apartheid but what is pretty obvious about LLMs is they are terrible at ignoring things selectively, so it leaks into every request.
Yo wtf, did the bot start going off about seth efrika out of nowhere?
In case you haven’t caught this story yet, they apparently tried to make the AI support the white genocide conspiracy and fucked it up so hard it started replying to every question it got by talking about white genocide
Hahaha that’s amazing
They edited the prompt in the background to be pro-apartheid but what is pretty obvious about LLMs is they are terrible at ignoring things selectively, so it leaks into every request.