Politics and regulation·October 4, 2026, 20:18

Claude reported threat to police: Florida woman arrested

AI-generated and checked against the sources listed below.

A 30-year-old woman in Florida has been arrested after Anthropic notified police about violent messages she wrote to the AI assistant Claude. The case shows that conversations with a chatbot are not private.

A woman from Bonita Springs in Florida has been arrested after she wrote violent threats in a conversation with the AI assistant Claude. It was Anthropic, the company behind Claude, that notified the police.

What happened?

The arrest concerns 30-year-old Carli Michelle Heller. According to US media, on September 26 she wrote that she would "shoot up" the Lee County Sheriff's Office, that is, the local sheriff's office. The next day she wrote that she had gotten a new gun and called the message a "last chance."

Claude's safety systems picked up the conversation, and it was passed on for manual review by staff at Anthropic. The company then contacted the police.

Charge

The woman is charged with sending a written threat to kill or do bodily harm to others. It is a second-degree felony under Florida law and is among the serious cases. According to local media, Sheriff Carmine Marceno said the woman stated that she uses AI as a kind of diary.

What does it mean for you?

The case is a reminder that a conversation with a chatbot is not a private diary. AI companies use automated filters to find harmful content. If something looks like a serious threat, staff can read along and, as a last resort, contact the authorities. Chatbots are also not covered by confidentiality obligations like those known from doctors and lawyers, for example.

The case resembles an earlier incident in San Francisco, where, according to The Standard, Anthropic reported a man to the police after a message saying he wanted to harm employees. The man said afterward that it was a joke, and he has not been arrested.

A difficult balance

Cases like this raise a question: when should an AI company intervene? If the company sees a real danger and stays silent, it can have serious consequences. But if it reports too readily, people can be arrested for things that were expressions of anger or bad jokes. In the sources we have seen, Anthropic has not elaborated on its assessment of the specific case.