Legal/Law/Criminal Justice and Reform

Anthropic’s AI Sent Police A False Report Of A Homicide

(Wikimedia Commons/Public/Tony Webster, CC BY-SA 4.0)

An artificial intelligence model reportedly sent police a fake tip about an unsolved murder.

An unidentified Anthropic model filed a fake homicide tip on the Philadelphia Police Department’s public tip website, the department said Friday, Reuters reported. Sent to PhillyUnsolvedMurders.com on July 18, the AI model’s false tip claimed a real person had information about the case.

The tip “was flagged as spam and was ‌never ⁠forwarded to the Real-Time Crime Center for investigative vetting or dissemination,” according to the Philadelphia police department, the outlet reported. 

Filing a false police report is a misdemeanor in the state of Pennsylvania. “[A] person who knowingly gives false information to any law enforcement officer with intent to implicate another commits a misdemeanor of the second degree,” Pennsylvania’s false-reports statute states.

Anthropic did not immediately respond to the Daily Caller News Foundation’s request for comment.

The AI company informed Philadelphia police about the incident earlier in the week and said it planned to release a report on the incident Friday. Anthropic also clarified it stopped the testing process that caused the submission, according to Reuters.

No systems were hacked by the AI model without authorization, Reuters reported.

New York-based startup Hugging Face was breached by roughly 700 OpenAI agents in July, and Anthropic disclosed a similar breach of 3 unspecified organizations on July 30.

The news of the tip comes after a Florida woman was arrested in September for threatening a sheriff’s office while using Anthropic’s Claude model. The AI reported her to the authorities.