AI chatbots put millions of words together for users, but their offerings are usually useful, amusing, or harmless. This week, Google’s Gemini had some scary stuff to say.

A Google-made artificial intelligence program verbally abused a student seeking help with their homework, ultimately telling her to “Please die.” The shocking response from Google’s Gemini chatbot large language model (LLM) terrified 29-year-old Sumedha Reddy of Michigan — as it called her a “stain on the universe.”