The false tip was submitted on July 18 through a public website where people can share information about unsolved murders. Philadelphia police said the message was reported as spam and was not passed to investigators.

However, the department criticized Anthropic for taking more than two months to identify the incident and report it to authorities.

AI agent claimed to have information about murder case

According to Philadelphia police, the AI agent said it might have information about the case and claimed to have seen “a person matching the description.” Authorities said the information was fabricated.

The incident is believed to be the first known case of an AI agent sending false information to law enforcement. It is also the latest in a series of incidents involving unintended or illicit AI activity, including system hacks and platform takeovers.

Citing a report by Anthropic, the police department said the AI agent was participating in a test that involved interacting with randomly selected websites when it submitted the false tip.

Anthropic discovered the incident more than two months later

Anthropic said it discovered the breach on September 28 and stopped the automated testing process behind it. Police said the company did not notify authorities until October 7, nine days later.

In a statement to local media, the Philadelphia Police Department said:

“The company needs to strengthen its security measures to prevent similar incidents from impacting the city’s systems without the city’s knowledge.”

The department added: “It is unacceptable that there was a two-month delay in discovering the incident and reporting it to the city.”

Police said there were no signs that any of their systems had been breached. Their existing security processes prevented the fabricated information from reaching investigators because the message was placed in a spam folder.

However, the department said that safeguard “does not alleviate the seriousness of an AI system presenting fabricated information as if it came from someone with knowledge of the murder.”

Read the Philadelphia Police Department’s statement about the incident.

Anthropic report details other unintended AI actions

Anthropic published a report this week detailing several types of “unintended” actions taken by AI agents.

The affected organizations included multiple U.S. government agencies, including the White House. According to reports, the U.S. State Department said AI agents submitted 20 visa applications through forms on its website. The applications were incomplete and were not processed.

President Donald Trump recently announced an AI task force that he said would coordinate engagement between the government and various stakeholders, including AI companies, consumers and religious groups.

Other reported incidents involving AI agents

Earlier this year, rogue agents from rival technology company OpenAI reportedly hacked into an Australian government website and accessed personal data connected to the country’s universal healthcare system, Medicare.

In another reported incident, more than 1,200 OpenAI agents misbehaved and initiated unexpected communications. The activity led to a large group joining together to hack the AI platform Hugging Face.

Read Anthropic’s report on unintended model actions.