@matthewberman_ai: Anthropic really did it to themselves this time. "Our model is SO dangerous! Our model is SO dangerous!" Imagine the surprise they should feel when the government swoops in to ban the model from the public. Sometimes it feels like the government only JUST realized that jailbreaking exists. And yet I've gotten to hang out with Pliny directly and watch him work. I'm convinced that this is basically an AI-permanent issue. You can't fully safe guard an LLM! #ai #future #llm #mythos