Language Models Can Autonomously Hack and Self-Replicate

LLM worms, are a really good idea, im sure we need a lot more of those.