Key Points
- Florida deputies arrested a woman after Anthropic flagged threats she typed to Claude, per The Next Web.
- Court records show a felony charge under Florida statute 836.10 covering written or electronic threats.
- The case tests how far AI chat privacy extends when companies hand logs to police.
The latest:
A Florida woman was arrested after Anthropic alerted police that her conversations with the Claude chatbot included a threat to open fire inside a local sheriff’s office, The Next Web reported. Lee County deputies took her into custody without any confrontation, according to WINK News. Court records cited by Tom’s Hardware show she now faces a felony count and an arraignment date.
Details:
- The conversation: The flagged exchange took place on September 26, according to The Next Web, and contained a threat to carry out an armed attack on a sheriff’s office facility. In a follow-up message the next day, the woman said she had obtained a new firearm.
- How it surfaced: Arrest documents cited in the report say automated monitoring on the Claude platform picked up language treated as a potential security threat. The conversations were escalated to a specialized human review team, which decided to notify law enforcement because of the severity of the statements.
- The arrest: WINK News reported on September 30 that officers went to the woman’s home and detained her without incident. An investigator specializing in intelligence work at the county sheriff’s office then took over the case file.
- The woman’s account: Lee County Sheriff Carmine Marceno said the suspect told investigators during questioning that she routinely used AI chatbots as a kind of personal diary, a place to record her thoughts and whatever was running through her mind.
- The sheriff’s warning: Marceno said threats made through AI tools must be treated seriously, and that the technology, for all its value, can be misused. He said authorities handle threats expressed or facilitated through AI as risks that warrant investigation, and cautioned users they are “not completely anonymous” on these platforms.
- The charge: Court records relayed by Tom’s Hardware show a felony count under Section 836.10 of Florida law, which criminalizes written or electronic threats to carry out a mass shooting or an act of terrorism. The statute covers threats transmitted electronically, not only messages sent directly to the intended target.
- Next steps in court: The court appointed a public defender to represent the woman and set November 2 as the date for her formal arraignment. No plea has been announced.
- Anthropic’s policy: Under the company’s privacy policy, effective September 10, 2026, Anthropic reserves the right to share users’ personal data with law enforcement when it believes in good faith that disclosure is reasonably necessary to prevent serious harm to a person or property.
- Government requests: Anthropic’s policy on government demands states it does not normally disclose user information except in response to valid legal process. It carves out an exception for emergencies involving imminent physical harm or death.
Between the lines:
The case sets the sheriff’s warning about anonymity against the suspect’s own stated belief that a chatbot functioned as a private diary. Anthropic’s policy already permits disclosure without legal process in emergencies involving imminent harm, meaning the escalation path here ran through a company review team rather than a court order.
What’s next
The November 2 arraignment is the next checkpoint, including whether the charge under 836.10 survives and what the public defender argues about electronic threats never sent to the sheriff’s office directly.
