A woman in Bonita Springs, Florida, has been arrested and charged with a felony after an artificial intelligence (AI) chatbot, Claude, developed by Anthropic, detected a violent threat in her personal entries. Carli Michelle Heller is facing a charge of making a written threat of violence under Florida law after the AI system flagged her alleged intent to “shoot up” the Lee County Sheriff’s Office.
Heller reportedly used Claude as a digital diary, confiding her thoughts and plans to the chatbot. On September 26, Claude’s internal safety systems identified the concerning entry and escalated it for human review. Upon determining the statement constituted a credible threat, the human reviewer promptly reported it to law enforcement authorities.
Anthropic, the company behind Claude, maintains a policy that allows for the sharing of user information in limited emergency situations, particularly when disclosure is deemed necessary to prevent death or serious physical injury.
Following the report, deputies identified Heller and proceeded to her home, where she was detained without incident. An intelligence detective from the Lee County Sheriff’s Office subsequently took over the investigation. Heller’s charge falls under Florida Statute 836.10, which designates sending, posting, or transmitting a written or electronic record threatening to kill, injure, commit a mass shooting, or carry out an act of terrorism as a second-degree felony, provided the communication is viewable by another person.
This incident highlights growing scrutiny over how AI companies manage potentially dangerous user interactions. Last month, reports surfaced regarding OpenAI and its CEO Sam Altman being sued in British Columbia over claims the company could have prevented a mass shooting. The alleged shooter had previously engaged in conversations about gun violence flagged by OpenAI’s safety team, though the company did not alert police as the discussions did not meet their threshold for legal referral.
Similarly, Florida has sued OpenAI and Altman, alleging that ChatGPT contributed to real-world harms, including a 2025 shooting at Florida State University. These cases underscore the complex ethical and legal challenges AI developers face in balancing user privacy with public safety, especially as human reviewers increasingly monitor AI conversations and prompts for potentially harmful content.