AI startup Anthropic reports that cybercriminals used its Claude AI chatbot for “vibe hacking” schemes that automate attacks, calculate ransom fees, and generate “visually alarming ransom notes.”