Anthropic Sent a Claude Diary Entry to Police on a Human Reviewer's Call, and a Florida Felony Charge Followed
A.I. / news
Anthropic Sent a Claude Diary Entry to Police on a Human Reviewer's Call, and a Florida Felony Charge Followed
Reports cite Anthropic's emergency-disclosure policy, not a statement on the case, and the cases reaching police since August now number at least three.
Anthropic reported a Claude conversation to police after its safety systems flagged a September 26, 2026 entry in which a Bonita Springs, Florida woman wrote that she would "shoot up" the Lee County Sheriff's Office. She now faces a charge under Florida Statute 836.10, a second-degree felony.
The account comes from TechSpot and from Tom's Hardware, as syndicated by Yahoo. Business Today carries the same sequence. The arrest report itself was not among the sources read, and the woman is not named here because the case has not reached court.
What the reports say happened
TechSpot reports that the woman uses Claude "like a diary". Claude's safety systems flagged the September 26 entry and escalated it to a human reviewer, and after the reviewer judged it a credible threat, Anthropic reported it to law enforcement.
Tom's Hardware reports that a message the next day allegedly mentioned a new gun. Deputies visited her home and detained her without incident, according to TechSpot. An arraignment was scheduled for November 2, and no public defender was listed on the docket at first, Tom's Hardware reports.
Anthropic's policy on emergency disclosure
The reports cite Anthropic's policy rather than a statement on this case. TechSpot quotes the company as saying it "may share user information in limited emergencies if it believes disclosure is necessary to prevent death or serious physical injury". Tom's Hardware quotes the privacy policy as permitting disclosure on a "good-faith belief that disclosure is reasonably necessary to prevent serious harm".
The two wordings differ, and neither report says which text Anthropic's policy page carries today. None of the reports cited here includes comment from Anthropic about the Lee County entry, how often its reviewers escalate conversations, or what threshold applies.
Lee County Sheriff Carmine Marceno warned that conversations with AI are "never truly anonymous", Tom's Hardware reports.
Third such case since August
Tom's Hardware counts this as at least the third case since August in which a conversation with Claude or ChatGPT reached police.
| Case | What was written | Outcome reported |
|---|---|---|
| Lee County, Florida | A diary entry about shooting up the sheriff's office | Felony charge, arraignment set for November 2 |
| San Antonio | Queries about shooting an elementary school | User faced charges |
| San Francisco | A threat against Anthropic CEO Dario Amodei | Outcome not reported |
TechSpot places the case against lawsuits brought against OpenAI by British Columbia and Florida over conversations about gun violence that were allegedly not reported. The direction of the pressure matters here: those suits allege too little reporting, and this case shows the opposite outcome.
Related coverage of consumer AI products and user data: the ChatGPT image-ad test and Agent Reach.
The next dated event is the November 2 arraignment in Lee County.
Sources
More in A.I.
- 01Reflection Announces Beam, a 501B Open-Weight Model, but Has Not Named a LicenceThe weights are promised for later in October. Reflection's own post gives scores and training scale, and says nothing on licence terms or API pricing.
- 02Meta Disputes Inc. Columnist's Claim That Muse Read His Messages While Mac Researcher Calls It a BackdoorTwo separate Muse problems are being reported as one. A researcher's local-access flaw is documented; the claim that the agent read private messages is contested by Meta.
- 03Anthropic Staff Reported a Claude Chat to Police, and a Florida Woman Faces a Felony ChargeThe company's privacy policy allows disclosure to prevent serious harm. Its transparency report counted zero emergency requests from police, and does not count referrals it makes itself.
- 04Vals AI's 90 Claude Opus 5.5 Agents Name Two Magnetic Semiconductor Candidates, Neither Yet MeasuredOne candidate was first synthesised in 1999 and has a measured ordering temperature of 376 K. The other may not survive the furnace.