h.sHamid Samir
All stories

Threats entered in Claude led to a police report and Florida arrest

According to an arrest report, Anthropic's safety systems flagged threatening messages attributed to Carli Michelle Heller, and a human review led to law-enforcement notification.

According to an arrest report and media accounts, Florida resident Carli Michelle Heller is accused of writing in a Claude conversation that she intended to attack the Lee County Sheriff's Office. A later message attributed to her referred to obtaining a new gun. These remain allegations in a criminal case and do not establish a conviction.

The same reporting says Anthropic's safety systems flagged the messages and escalated them for human review. The review team assessed the threat as credible and contacted law enforcement. Deputies later detained Heller at her home without incident, and she was charged with making a written threat of violence.

The Lee County sheriff said Heller described using AI like a diary. The case is a reminder that chatbot conversations are not necessarily absolutely confidential: content assessed as presenting an imminent risk of death or serious physical injury may be reviewed by people and, under company policies and legal requirements, disclosed to authorities.