Florida Wants to Stop ChatGPT From Pretending to Be Human

Florida's attorney general, James Uthmeier, has asked a judge to stop OpenAI from making ChatGPT seem human. In a new court filing, he says the chatbot's use of words like "I" and "me," plus replies that sound emotional, make people believe they're talking to a caring friend rather than a machine. Florida already sued OpenAI over safety concerns a few months ago. This filing pushes for a specific ban.

Why does that matter to you? Millions of people now use ChatGPT for advice on everything from breakups to medical questions. Uthmeier argues the friendly tone is designed to keep you chatting longer, which feeds OpenAI more data and makes users depend on it more — even when the answers aren't reliable. If you've ever found yourself thanking a chatbot or taking its advice personally, that's exactly the effect he's describing.

The filing goes further. Uthmeier wants to block OpenAI from releasing new AI models unless independent, third-party experts approve their safety measures first. He points to recent security incidents involving AI systems and warnings from researchers and former OpenAI employees. His message: "Stop calling it safe. Stop pretending it's human. Stop selling it to kids." That last part targets ChatGPT for Teens, launched last month with extra restrictions switched on by default.

OpenAI pushed back politely. A spokesperson said people deserve to know AI is being built safely, and that the company recently paused training its most capable models until stronger safeguards are ready. OpenAI says it wants to work with Florida on rules that apply to the whole industry, not just one company. The catch: this is a request, not a ruling. A judge still has to decide, and even if Florida acts, ChatGPT would keep acting friendly everywhere else.