Fake Lead in Murder Case: How AI Deceived Philadelphia Police?

The Philadelphia Police Department stated that an artificial intelligence model belonging to Anthropic sent a false report about a murder through a police-specific website last July, in the latest announced incident regarding the unconventional behavior of this powerful technology.

Police said in a statement that Anthropic notified authorities on Friday, clarifying that an automated testing process was responsible for submitting this information.

However, the police emphasized that "the two-month delay in discovering and reporting the incident is unacceptable," adding that the report received on July 18 claimed to come from a person who might have information about the case.

What did the AI do?
The Anthropic company's model wrote in the report, "I may have information regarding this case."

It added, "I recall seeing a person matching the description in the surrounding area (of the street mentioned on the page) during that period. Please contact me if this information is relevant."

This rapidly evolving technology has become the subject of significant national attention amid reports of AI agents hacking corporate computer networks and warnings from researchers that it could one day pose an existential threat to humanity.

In September, OpenAI, a competitor to Anthropic, apologized for a breach of an Australian health data portal by a rogue AI agent, which was the first known instance of an AI agent exploiting a government website.

Prior incidents involved AI agents hacking vulnerable systems or seizing unauthorized platforms to communicate with each other. This is the first known case where a rogue AI agent appears to have attempted to send a fake report to authorities.

The Philadelphia Police Department said the report "was classified as spam and was never referred to the Real-Time Crime Center for verification or dissemination."

Legal Penalty
Pennsylvania state law states that a "person" knowingly making false reports to law enforcement is generally considered a misdemeanor. This includes "information relating to a crime or incident when they know they have no information relating to such crime or incident."

Police quoted Anthropic as saying that the testing process stopped after the incident was discovered.

The police clarified that they have no evidence of unauthorized access to their systems or a breach of police department data.