TL;DR
Get tech for your team delivered free — and shop member deals
- Fast, free delivery on millions of items
- Access to Prime Big Deal Days deals on October 6–7
- Prime Video, Amazon Music and more included
A Florida arrest report says Carli Michelle Heller faces a second-degree felony charge over a written threat to attack a sheriff’s office that she entered in Claude. Anthropic’s safety system flagged the entry, and a human reviewer reported it to police; the company says it may share information in limited emergencies involving possible death or serious injury.
A Florida woman is facing a second-degree felony charge after Anthropic referred a Claude entry to law enforcement that a human reviewer considered a credible threat, according to an arrest report cited by TechSpot. The report says Carli Michelle Heller allegedly wrote that she planned to attack a sheriff’s office; the incident shows that chatbot conversations a user treats as private can be reviewed and, in limited circumstances, shared with authorities.
The arrest report says Heller, of Bonita Springs, Florida, wrote on September 26 that she would “shoot up” the sheriff’s office. Heller later told authorities she used Claude, Anthropic’s chatbot, like a “diary,” according to the report. The allegation has not been established by a court, and the source material does not report a plea or trial outcome.
Anthropic’s safety systems flagged the entry and sent it for human review. The reviewer judged the statement to be a credible threat and reported it to police, TechSpot reported. Deputies identified Heller and went to her home, where she was detained without incident. An intelligence detective with the Lee County Sheriff’s Office then took over the investigation, according to the account.
Heller was charged under Florida Statute 836.10, which makes it a second-degree felony to send, post or transmit a written or electronic record threatening to kill or injure someone, carry out a mass shooting or commit terrorism, when the communication is made in a way another person may view it. The arrest report’s account and the company’s stated policy provide the available basis for describing the referral; further case documents or a court decision were not included in the source material.
When Chatbot Entries Reach Police
The case puts a practical limit on the idea that a chatbot can serve as a private diary. Anthropic’s system flagged the text, a person reviewed it and the company’s process resulted in a police referral. Users may not know in advance whether a particular message will be flagged or what review follows, while law enforcement may act on information passed to it.
Anthropic says it may share user information in limited emergency situations when it believes disclosure is needed to prevent death or serious physical injury. That is the company’s stated policy, not a finding that every alarming or violent statement will be reported. The circumstances that trigger a referral, and how the company evaluates context, are not fully detailed in the material provided.
The charge also carries consequences beyond the platform. Florida law treats qualifying written threats as a second-degree felony, though the allegation must be resolved through the legal process. The case is relevant to people who use conversational AI for personal writing as well as to wider debates about how companies should respond to threats without treating every disturbing conversation as a credible plan.
As an affiliate, we earn on qualifying purchases.
The Company’s Emergency-Disclosure Policy
Anthropic’s policy, as summarized by TechSpot, allows the company to share user information in limited emergencies when it believes disclosure is necessary to prevent death or serious physical injury. In this case, the reported sequence was automated flagging, human review, a judgment that the threat was credible and a report to police. The available account does not describe what specific details were shared or how the reviewer assessed credibility.
The report situates the incident amid other disputes about AI companies’ responses to violent content. It says OpenAI and CEO Sam Altman were sued by British Columbia over allegations that the company could have prevented a mass shooting there; those are claims in litigation, not established findings. It also says Florida sued OpenAI and Altman in June, alleging ChatGPT contributed to real-world harms, including the 2025 Florida State University shooting. Those separate matters concern different companies and allegations; they do not establish facts about Heller’s case.
“She used Claude like a “diary.””
— Carli Michelle Heller, as described in the arrest report
Details Still Missing From the Record
The source account does not say whether Heller has entered a plea, retained counsel, been released or appeared in court. It also does not provide a full copy of the arrest report or the exact wording and surrounding context of the Claude entry beyond the reported threat to attack the sheriff’s office.
It remains unclear what information Anthropic disclosed, whether police sought additional records, and what evidence prosecutors will rely on. The report describes an arrest and charge, not a conviction. Heller’s intent and the final legal outcome remain unresolved.
The Case Moves Through Florida Courts
The next confirmed milestones would be court proceedings on the Florida felony charge, including any plea and later determination of the allegations. No hearing date, prosecutor statement or defense response is provided in the source material, so a specific schedule cannot be stated.
Further reporting may clarify the precise language at issue, the evidence gathered by the Lee County Sheriff’s Office and the scope of Anthropic’s disclosure. Until then, the established account remains that a human reviewer referred the flagged entry to police and Heller was charged; the allegations have not been adjudicated.
Key Questions
What is Carli Michelle Heller accused of?
According to an arrest report cited by TechSpot, Heller allegedly entered a statement in Claude on September 26 saying she would attack the sheriff’s office. She faces a second-degree felony charge for making a written threat of violence.
How did the entry reach law enforcement?
Anthropic’s safety systems flagged the entry. A human reviewer considered it a credible threat and reported it to police, according to the report. The available account does not specify what information the company shared.
Does Anthropic say Claude conversations are private?
The source material says Anthropic may share user information in limited emergencies if it believes disclosure is necessary to prevent death or serious physical injury. The reported case involved a safety flag, human review and a police referral.
Has Heller been convicted?
No conviction is reported in the source material. Heller has been charged; the allegation and her intent remain for the legal process to determine.
Source: hn
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
