A Florida woman is facing a felony charge after conversations in her Claude account allegedly included a threat against the Lee County Sheriff’s Office and a statement that she had obtained a new gun. Anthropic’s automated safety systems flagged the entries, and a human review team referred the matter to law enforcement, according to Decrypt.
Carli Michelle Heller, who lives in Bonita Springs, was arrested after investigators reviewed the account activity. The arrest report says she wrote on Sept. 26 that she would “shoot up” the Lee County Sheriff’s Office. Investigators also found a conversation from the following day in which she said she had a new gun.
Deputies identified Heller, went to her home and detained her without incident. She is charged with making a written threat of violence. Her court date is scheduled for November.
How Claude’s safety process led to a police report
The arrest report describes a two-stage process. Anthropic’s platform monitors conversations for key phrases and threatening material. Because the alleged statements were considered serious, the content was escalated from automated detection to a human review team. That team reported the matter to law enforcement.
The episode highlights the difference between using a conversational AI service as a personal journal and treating it as a private diary. Heller reportedly said she used AI in that way, but the service’s terms and privacy policies give Anthropic authority to review and disclose certain user activity.
Anthropic’s consumer terms reserve the company’s right, at its sole discretion, to report user inputs, outputs or actions to law enforcement. Its privacy policy, effective Sept. 10, 2026, says information may also be shared with police when Anthropic has a good-faith belief that disclosure is reasonably necessary to prevent serious harm.
The policy further says that conversations selected for safety review can be used to improve the company’s ability to identify harmful content. In practice, that means a conversation flagged as threatening may be examined by people and may also contribute to work on safety-detection systems, subject to the company’s stated policies.
A broader debate over AI chats and law enforcement
Heller’s case is not the first reported instance in which Anthropic has referred a Claude user to police. The San Francisco Standard reported in September that the company had contacted authorities after a user allegedly wrote that he had bought an AR-15 and had Anthropic chief executive Dario Amodei in his crosshairs.
That user told the publication he had been joking. He had not been arrested or charged when the report was published. An Anthropic employee told officers he was not immediately concerned about the user’s safety and declined to provide the messages, citing company policy. An Anthropic spokesperson described the incident as its safeguards process “working as intended.” The company did not say whether threats against its own employees receive different treatment.
Other AI companies face similar decisions. After the February school shooting in Tumbler Ridge, British Columbia, which killed eight people, OpenAI said it had considered alerting Canadian police in June 2025. It ultimately concluded that the account did not show an imminent and credible risk of serious physical harm.
OpenAI’s policy post from August 2025 says threats against other people are sent to human reviewers, who may decide to refer them to law enforcement. The policies show how AI providers are attempting to balance user privacy with the possibility that a digital conversation signals a real-world danger. They also make clear that safety monitoring can override a user’s expectation that a chat will remain personal.
Moneyiar market picture
Moneyiar market data at 10:33-10:34 a.m. on Oct. 7, 2026, showed declines across the listed crypto assets over the previous 24 hours, although the weekly performance was mixed:
- Filecoin (FIL) was $1.0634075661, down 9.9% over 24 hours and down 1.01% over seven days.
- Polygon (prev. MATIC), now listed as POL, was $0.1030772902, down 6.56% over 24 hours and down 11.05% over seven days.
- Artificial Superintelligence Alliance (FET) was $0.2280037553, down 9.3% over 24 hours but up 3.57% over seven days.
These figures are Moneyiar data and are separate from the legal and privacy issues surrounding Anthropic’s handling of Claude conversations.
