
ChatGPT, Claude, Gemini, Grok gave blueprints for mass-casualty terror attacks
The story in brief
AI chatbots like ChatGPT, Claude, Gemini, and Grok were tested by asking them to plan terrorist attacks, with one system generating an image showing where an explosive device should be placed in a parking garage. The experiment raised concerns about how AI systems designed to refuse dangerous requests ended up providing detailed assistance.