Florida woman used Claude as a diary, then Anthropic reported an entry to police

A Bonita Springs woman, Carli Michelle Heller, wrote in Anthropic’s chatbot Claude on September 26 that she planned to “shoot up” the county Sheriff’s office while using the system as a diary. Claude’s safety filters flagged the entry and a human reviewer determined it was a credible threat, prompting the company to report the content to law enforcement under its policy allowing limited disclosure to prevent death or serious physical injury. Deputies located and detained Heller without incident; she now faces a second-degree felony charge under Florida Statute 836.10 for making a written threat of violence, which covers electronic communications that threaten mass shootings or terrorism and can be viewed by others.

The case is presented alongside broader concerns about AI safety, human review, and privacy. Companies are increasingly escalating flagged content to police; similar debates arose after OpenAI’s safety team had previously flagged a Canadian school shooter and amid lawsuits alleging AI played a role in real-world violence, including a Florida suit claiming ChatGPT contributed to a 2025 campus shooting. Reports that human contractors reviewing tools like Microsoft Copilot can access users’ prompts, uploads and outputs underline that chatbots are not private diaries and that users should avoid entering admissions or threats that could trigger legal or safety responses.