A 30-year-old Bonita Springs woman was arrested after typing two threatening messages into Anthropic’s Claude chatbot in late September 2026; automated filters flagged the texts, human reviewers escalated them to the FBI, and the FBI passed a tip to the Lee County sheriff. Messages included explicit threats to shoot the sheriff and referenced buying a gun; she says she used the assistant as a diary, admits writing the texts but denies intent to act or owning a weapon, and has pleaded not guilty. The arrest was recorded under Florida Statute 836.10, which criminalizes written or electronic threats viewable by others and carries up to 15 years in prison; prosecutors had not yet formally charged her as of early October and an arraignment was set for November 2.
Anthropic’s policies and practices matter here: automated classifiers flag content, humans review flagged chats, and the company’s consumer and European terms allow sharing information with law enforcement to prevent “serious harm,” with flagged conversations retained up to two years even if deleted by users. Legal scholars question whether a private chatbot entry counts as a “written threat” under Florida law, and similar incidents have led to other U.S. arrests this year. Businesses are advised to treat chatbots as non‑private, use contractual business plans, strip identifying data before pasting, and codify AI use rules, because reporting thresholds and cross‑platform attribution remain murky.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
The Consumer Financial Protection Bureau (CFPB) has launched an investigation into nine large national banks over alleged debanking practices. The inquiry aims to determine whether these banks have unlawfully restricted access to banking services. (forbes.com)
British Airways has updated its rules to prohibit covert filming of employees with Meta smart glasses, warning that doing so can lead to flight cancellation and refusal of service. The airline can also report incidents to authorities if recording without consent is suspected. (liveandletsfly.com)
The EPA has retracted its claim that the US has phased out the toxic PFAS compounds Pfoa and Pfos. Despite recognizing their dangers, the agency continues to permit production and use of these chemicals in certain products. (theguardian.com)
Namecheap has seized and disabled all domains associated with an account, causing disruption. The user reports being blocked and unable to access their domains or account features. (reddit.com)
Python's precompiled binary wheels simplify package installation but pose security risks because attackers can upload malicious binaries that do not match the source code. Analyzing and verifying binary integrity is difficult, making supply chain attacks increasingly likely in the Python ecosystem. (rushter.com)
Microsoft has withdrawn from a major AI datacentre project in Essex due to concerns over power supply stability. The project, led by UK firm Nscale, was expected to be the largest of its kind in the country. (theguardian.com)
Today's best Hacker News stories, summarized and screenshotted, one email a day.