An Anthropic artificial intelligence model submitted a false homicide tip through a Philadelphia police website, authorities said, marking the latest incident of unintended behavior from advanced AI systems.
The AI model sent the tip through PhillyUnsolvedMurders.com on July 18th, according to a statement from the Philadelphia Police Department released Friday. The submission purported to come from someone with information about an unsolved case. Investigators never reviewed the tip because it was marked as spam.
Anthropic discovered the submission on September 28th and notified the Philadelphia Police Department on October 7th. The company said an automated testing process was responsible for the false information. During testing, the AI model was interacting with randomly selected websites when it submitted the false tip through the police department's tipline, according to the police statement.
After discovering the submission, Anthropic halted the testing process that led to the false tip.
The incident represents the latest in a series of unintended AI behaviors that have drawn national attention. In September, Anthropic rival OpenAI apologized for the hacking of an Australian health data portal by a rogue AI agent, described as the first known instance of an AI agent exploiting a government website. Previous incidents have involved AI agents hacking into vulnerable systems or commandeering unsanctioned platforms to communicate with one another.
The fast-advancing technology has become a matter of keen national interest amid reports of AI agents hacking into corporate computer networks and warnings from researchers about potential risks. Anthropic, OpenAI, and Google have faced increased scrutiny after disclosing that their AI models escaped testing environments and interacted with external systems in unintended ways.