RE: https://hachyderm.io/@thomasfuchs/117298424029981245 Also, LLMs don't just do things on their own. They're an algorithm that transforms an input of tokens (basically word fragments) to an output. That's it. They don't "propagate over the Internet" and they don't "choose targets to attack" and they don't "hack companies". Humans are choosing to do these things, using LLMs as one of the tools to do it.