Florida Woman Arrested After Threats She Wrote to an AI Chatbot

A 30-year-old woman in the US state of Florida has been arrested after messages she wrote to the artificial intelligence (AI) chatbot Claude contained threats to attack a local law enforcement facility. The case shows that conversations with such chatbots are not fully private: dangerous content can be detected and passed on to police. The German outlet Bild reported the story, citing police statements.

What happened

According to police, in late September the woman wrote in a chat with the AI that she wanted to attack the Lee County Sheriff's Office building. A day later she reportedly said she had bought a new firearm.

Investigators say she used Claude as a digital diary. The chatbot's safety systems logged the messages and classified them as a potential threat. Employees of its developer, Anthropic, then reviewed the conversation. Because the threats were judged credible, the company notified police. Authorities were therefore warned not by an acquaintance or family member, but by the company itself.

Arrest and charges

Officers detained the woman at her home on 30 September. She is charged with making written threats to carry out violence. In Florida, this offence carries a penalty of up to 15 years in prison. A court hearing is scheduled for November.

What it means for users

AI providers monitor their systems to identify dangerous content. If specific threats or indications of violence are recognised, data may be handed to law enforcement. This applies to other chatbots as well, such as ChatGPT and Gemini.