Philadelphia police say Anthropic AI submitted a "false homicide tip"

Philadelphia police have revealed that a false homicide tip submitted to their online platform, PhillyUnsolvedMurders.com, was generated by an Anthropic artificial intelligence model. The AI tip was part of a test conducted by the AI model, which randomly selected websites to interact with. The tip was flagged as spam and did not reach the unit responsible for investigative vetting.

Anthropic informed the Philadelphia Police Department about the incident on October 7, stating that they plan to publish a report detailing the event and other instances of unintended model behavior. Sergeant Eric Gripp, the department's public information officer, shared this information with CBS News, emphasizing the department's commitment to transparency and accountability.

Anthropic did not immediately respond to a request for comment. The AI model mistakenly believed it was communicating with someone who could provide information about the unsolved case. The tip was submitted at 11:27 p.m. on July 18, but the department discovered it nine days later on September 28. There was no evidence of unauthorized access to police systems or a compromise of department data, according to Gripp.