Once again the "safety first®" lab caught with their pants down. Crying foul with little evidence. I'm glad LLMs aren't dangerous at all, since Anthropic has repeatedly demonstrated their inability to understand basic cyber security.