Anthropic reported diary entry to police, woman faces felony charge
TL;DR Highlight
Anthropic reported a Claude diary entry to police, and a woman now faces a felony charge.
Who Should Read
Service developers who process user conversations through LLM APIs and must set log retention and reporting policies
Core Mechanics
- The source article is missing
- This input has only comments and no article body.
- Facts about the article are limited to the title and what comments quote.
- What the title tells us
- Anthropic reported diary content to police.
- A woman faces a felony charge.
Community Reactions
- Opinion that the legal case is weak
- Many commenters quoted Florida Statute 836.10, which requires the threat to be sent or posted where another person may view it.
- Some argued a private chatbot diary does not meet that bar and she would win.
- Opinion that Anthropic had no good option
- One commenter cited headlines after OpenAI failed to report a shooter in a similar situation.
- They saw it as damned if you do, damned if you don't.
- Opinion that privacy expectations should drop
- One commenter said you are chatting with Big Tech, not a friend.
- Some suggested running open source models yourself.
- Experience that roleplay can't be told from intent
- One commenter said they tested LLMs and their company's internal flagging with extreme statements.
- They worried a human reviewer can't tell it isn't meant seriously.
How to Apply
- State your reporting policy in the terms of service
- Today, check whether your terms and privacy policy say when threatening conversations may be given to law enforcement.
- If not, draft wording with legal counsel.
- Audit log retention and access
- List who can read conversation logs.
- Shorten retention for raw logs you don't need.
Terminology
Related Papers
The Judge Can Debias Itself: Awareness and Self-Debiasing for Reliable LLM-as-a-Judge
Warning an LLM judge about bias or letting it self-critique improved 20 of 21 model–bias combinations.
90% of US working hours are spent on tasks not covered well by AI benchmarks
Mapping 29 benchmarks to job tasks shows 90% of US working hours lack good coverage
Postgres 19 delayed after AI reviewers find more bugs than humans can fix
LLM code review surfaced so many bugs that Postgres 19 slipped to November and 8-10 features were reverted
Show HN: Gemini 4 Argon plays StarCraft
LLMs wrote StarCraft bots in one hour; GPT-6 Astra scored highest.
A Plausible Explanation Is Enough to Mislead Multimodal Judges
Adding plausible explanations to fixed answers cut MLLM judges' wrong-answer rejection by 12.7–13.8 points.
Neural networks are clean
Looking at neurons directly shows most have one main job, challenging superposition.