A Florida woman is facing a felony charge after police say she wrote threats against a sheriff's office in conversations with Claude, the AI chatbot made by Anthropic.
Carli Michelle Heller, of Bonita Springs, said she used the AI like a diary. Those entries were flagged and passed to law enforcement, according to an arrest report cited by local outlet WINK News.
How the Case Unfolded
Investigators say Heller wrote on Sept. 26 that she would "shoot up" the Lee County Sheriff's Office. By the next day, her Claude account showed conversations saying she had gotten a new gun.
The arrest report says Anthropic's platform monitors for key phrases and threatening content. Because the statements were severe, they were sent to a human review team.
That team reported the messages to law enforcement. Deputies then identified Heller and went to her home.
She was detained without incident. Heller is charged with making a written threat of violence.
What Anthropic's Policies Allow
Anthropic's consumer terms say the company may report a user's inputs, outputs, or actions to law enforcement at its "sole discretion." This is stated openly in the company's terms.
Its privacy policy, which took effect Sept. 10, 2026, also allows sharing data with police. This applies when the company has a good-faith belief that disclosure is reasonably necessary to prevent serious harm.
The same policy says conversations flagged for safety review are used to improve Anthropic's ability to detect harmful content. This means a flagged entry can be used to train the company's detection systems.
Heller's case is not the first. In September, the San Francisco Standard reported that Anthropic referred a user to police after he allegedly wrote that he had bought an AR-15 and had CEO Dario Amodei in his crosshairs.
An employee told officers he was not immediately concerned about safety and declined to show the messages, citing company policy. The user told the Standard he was joking and had not been arrested or charged when the story ran.
An Anthropic spokesperson called the episode "our safeguards process working as intended." The company declined to say whether it treats threats against its own employees differently.
OpenAI has faced similar questions. After a February school shooting in Tumbler Ridge, British Columbia, that killed eight people, OpenAI said it had weighed alerting Canadian police in June 2025.
The company concluded at the time that the account did not show an imminent and credible risk of serious physical harm. In April, a lawsuit was filed in federal court in Northern California claiming OpenAI failed to warn police before the shooting.
OpenAI's August 2025 policy post says threats against others are sent to human reviewers. Those reviewers may refer cases to law enforcement.
Heller remains charged with making a written threat of violence. She has a court date set for November.
