Anthropic Reports Threat to Florida Sheriff's Office, Woman Arrested on Felony Charge
An Anthropic Claude conversation flagged by the company's safety systems led to the arrest of a Lee County woman on charges related to threatening a mass shooting at a local law enforcement facility.
Following a report submitted by Anthropic's content review team, a woman in Lee County, Florida, was taken into custody and charged with a felony after threatening language emerged in conversations with the Claude AI chatbot. According to WINK News, the individual identified as "Carli" made statements on September 26 indicating an intention to "shoot up" the Lee County Sheriff's Office, with investigators noting that a subsequent message referenced acquiring a firearm. This marks at least the third such incident in which threatening material from Claude conversations has been escalated to law enforcement since August.
Anthropic employs automated systems to scan conversations for language patterns and content suggesting potential threats, the arrest report indicates. When material reaches a certain threshold of concern, it can be forwarded to human reviewers within the company. In this instance, the review team determined the content warranted notification to authorities. Officers from the Bonita Springs area apprehended the suspect without resistance at her residence, after which an intelligence detective with the sheriff's office assumed investigative responsibility. While the formal arrest report has not yet entered the court system, public records confirm a felony charge was filed on September 30 under Florida Statute 836.10, which addresses written or electronic threats involving mass shootings or terrorist acts.