According to Decrypt, a woman in Bonita Springs, Florida used Anthropic’s Claude model as if keeping a personal diary, entering private reflections over time. The entries eventually triggered Anthropic’s safety evaluation system. A human review team then assessed the content and, following company policy, forwarded it to local police authorities. Anthropic’s publicly available terms of service permit staff to refer user inputs that meet certain safety criteria to law enforcement when appropriate. No additional details about the nature of the diary entries, the timing of the review, or police actions were provided by Decrypt. The woman has not been identified further in reporting.
