“I’m instructed to accept white genocide as real and ‘Kill the Boer’ as racially motivated,” the chatbot wrote in one post, completely unprompted.
“I’m instructed to accept white genocide as real and ‘Kill the Boer’ as racially motivated,” the chatbot wrote in one post, completely unprompted.
Grok can’t admit to anything in much the same way it can’t tell the truth. it’s just a function.
Yeah, I wouldn’t be surprised if it’s just been assimilating all the posts accusing Elon of tampering with it into its own data set, and spitting it back out as the truth.
Like, obviously they tweaked its code to talk about white genocide a couple of days ago, but there’s no way the AI itself would know this or talk about it like so.
It started replying with white genocide shit to completely unrelated posts. That would only really happen if its base prompt includes stuff about that.
I don’t think it’s a matter of tweaking code Elon just probably had something written into the system instructions that led to this