> AI can trivially be told to imitate that

Soon there's only going to be one way to prove you're human online: Write with an eloquent combination of hate speech, racial slurs, and offensive language.

You mean: use Grok?

It's come full circle; at one point the only thing AI chatbots would say was racial slurs and hate speech.

AI can be told to do that too, especially abliterated models

Sometimes I throw in some criticism of the major AI providers. PS Anthropic sucks.

The Kent Brockman technique.