Key Notes
- Investigators say alleged threats in Claude chats were flagged for human review and reported to police.
- A Florida woman faces a written-threat charge.
- Anthropic permits emergency disclosure to help prevent imminent harm.
A Florida woman was arrested after threatening messages she allegedly entered into Claude were flagged for human review and reported to law enforcement, according to investigators cited by local media. The case brings Anthropic’s safety monitoring and emergency disclosure policies into focus as people increasingly use chatbots for personal conversations.
Carli Michelle Heller, 30, of Bonita Springs, faces a charge of making a written threat of violence. Sheriff Carmine Marceno told WINK News she later described using AI like a diary. The allegations have not been established by a conviction.
How the Claude Conversation Reached Police
The arrest report, as reported by Gulf Coast News, describes a September 26 message threatening to attack the Lee County Sheriff’s Office. Investigators said a subsequent message mentioned obtaining a new gun. That reported statement does not independently establish that a weapon was purchased or recovered.
According to the investigators’ account, the platform flagged the material and escalated it to a human review team, which notified police. Deputies then went to Heller’s home and detained her without incident, WINK News reported.
That sequence matters: the reported disclosure followed human review. It does not establish that Claude independently contacted police, nor does it show what the assistant replied or explain the internal criteria used to escalate this particular conversation.
What Anthropic Says About Disclosure
Anthropic’s government request policy says it generally supplies customer information in response to valid legal process, such as a subpoena or warrant. It allows an exception when the company believes an emergency could cause imminent physical harm or death and immediate disclosure could help avert it.
The same policy says Anthropic evaluates requests for legal validity and may reject overly broad or vague demands. Users ordinarily receive notice when their data is requested, subject to legal restrictions and exceptions that include emergencies. These are general rules, rather than a case-specific explanation of the Florida disclosure.
Its privacy policy also permits sharing personal data with authorities when the company has a good-faith belief that doing so is reasonably necessary to prevent serious harm, address unlawful activity or protect safety. The policy states that Anthropic collects user inputs and outputs.
Why the Criminal Charge Is Separate From the Safety Review
Florida’s written threats law covers written or electronic records containing threats to kill, cause bodily harm, conduct a mass shooting or commit terrorism, when transmitted in a way another person may view. A violation is classified as a second-degree felony.
Whether the evidence satisfies that law is a question for the legal process. A safety team’s decision to escalate a message and a criminal finding are separate judgments. Describing a chatbot as a diary does not resolve how a court will assess the alleged statements or their context.
A Conversational Interface Still Has Data Rules
Anthropic’s transparency hub says its Safeguards Team develops detection and monitoring systems to enforce usage policies. It also describes handling law-enforcement requests for conversation content, account records and emergency disclosures. Those published practices make clear that a conversation can become part of a platform safety process.
We recently covered a different aspect of Claude’s enforcement policies involving regional access. The Florida case concerns alleged threats rather than where an account is used, but both raise questions about how a consumer service explains its rules to users.
For people using AI to organize personal thoughts, the relevant distinction is between a conversation feeling private and the provider’s actual commitments. The Florida allegations illustrate a reported escalation involving threats of violence; they do not establish that ordinary diary entries routinely lead to police reports.
Disclaimer: AIstify is an independent media brand owned and operated by NuvexMedia LLC, publishing news, research, and insights on artificial intelligence, emerging technologies, automation, and related industries. NuvexMedia LLC invests in and collaborates with companies across the AI, technology, software, and digital innovation sectors. These relationships do not influence AIstify’s editorial coverage, and the publication maintains full editorial independence to provide accurate, timely, and objective information. © 2026 NuvexMedia LLC. All rights reserved. This content is for informational purposes only and should not be considered legal, tax, investment, financial, or other professional advice.
