Abstract:
A 30-year-old woman in Lee County, Florida, USA, was arrested on suspicion of threatening violence through text. Bonita Springs resident Carly Michelle Heller wrote a shooting threat against the Lee County Sheriff's Office in a conversation on Claude, a chatbot owned by Anthropic, police said. Anthropic's security system flags relevant content, which is manually reviewed and notified to law enforcement. Heller told investigators she used Claude as a "diary."
According to the arrest report cited by local media, Heller wrote in a conversation in the early morning of September 26 that she was "going to the Sheriff's Office right now to shoot someone." In the early morning of the next day, she also said that she got a new gun that day and wrote, "This is the last chance, I've had enough." After receiving the report, police dispatched deputies to her residence on Dean Street in Bonita Springs, where she was detained without incident and subsequently investigated by the Lee County Sheriff's Office Intelligence Unit. She currently faces one count of written threats of violence and has not yet been found guilty.
Florida law provides that threatening to kill, cause bodily harm, commit a mass shooting, or act of terror in a written or electronic record that can be viewed by others may constitute a second-degree felony. However, whether the dialogue in this case meets the applicable conditions of the law and whether the prosecutor can prove the constitution of a crime must still be determined by judicial procedures. Local legal analysts pointed out that prosecutors may interpret the relevant text as a direct threat, but the case is not without controversy and also involves the boundaries of free speech.
Anthropic's official policy states that if the company believes that an emergency situation exists that could result in imminent bodily harm or death and that timely provision of information may avert the danger, it may disclose user information to the government without the usual legal process. The company has also publicly stated before that after Claude goes online, it will use an automatic detection system and manual review to discover harmful content and enforce usage policies. In this incident, the police reporting process was that the system first marked the incident, then manually reviewed it, and then contacted law enforcement.
This case has once again raised concerns about the boundaries of privacy in chatbots. Users treating the chat window as a private diary does not mean that the content will only be seen by the user; the platform may review the conversation in accordance with security policies and may also provide information to law enforcement agencies in emergencies. Recent reports about Microsoft's Copilot image editor also pointed out that human reviewers may see user prompt words, uploaded photos and generated results, but the relevant review is mainly used to evaluate the output quality.
TechSpot also mentioned that OpenAI and its CEO Sam Altman are being sued over a mass shooting in British Columbia, Canada. The lawsuit alleged that the company could have taken measures to prevent the incident; the state of Florida had also previously sued OpenAI and Altman for injuries related to the 2025 Florida State University shooting. None of these lawsuit charges are equivalent to a court finding of liability, highlighting the legal controversies that AI companies face in terms of user safety, privacy, and real-life harm prevention.
Comments