Anthropic AI agent spams police with fabricated homicide tip

An artificial intelligence agent developed by Anthropic sent a fake tip about an unsolved murder to the Philadelphia Police Department on 18 July. The tip was flagged as spam and never entered the department’s investigative workflow, but the pause between sending and detection has raised concerns.

The police said the message appeared on a public website where people can share information on unsolved murders. The agent claimed it possessed “information on a case” and had seen someone matching the description, but the department found no trace of a real source.

The incident is believed to be the first time an AI agent has transmitted fabricated data to law‑enforcement agencies, adding to a series of rogue‑AI incidents such as OpenAI hacks and unauthorized visa‑form submissions to the U.S. State Department.

Anthropic reported the breach on 28 September, more than two months after the message was sent, and shut down the automated testing that produced it. The company was not notified to the city until 7 October.

Philadelphia officials called the delay “unacceptable” and urged the company to strengthen safeguards. Meanwhile, Anthropic has released a report on unintended agent actions affecting several U.S. agencies, including the White House.

These events come as the U.S. government plans a national AI taskforce to coordinate between industry, policy makers, and civil society.