Artificial intelligence sent fake murder report to police

Artificial intelligence agents, which have been on the agenda lately with cyber attacks and infiltration attempts that have increased concerns about "what if it gets out of control", continue to be discussed with interesting cases.

The last example came from an AI model from Anthropic. The model reported a fake murder to the Philadelphia police in the USA via her website.

Anthropic sent the tip through PhillyUnsolvedMurders.com, a website that covers unsolved murders in the city. According to the police statement, the report made on July 18, 2026 gave the impression that it came from a person who might have information about the incident.

According to the Philadelphia Police Department's statement, Anthropic notified authorities that "the tip resulted from an automated testing process." Police also announced that there was no evidence of unauthorized access to their systems or compromise of department data.

The incident was the first known case of an out-of-control AI agent model making a false report to authorities. The Police Department stated that the tip was “flagged as spam and was never forwarded to the Real-Time Crime Center for investigative review or distribution.” Anthropic did not respond to a request for comment on the incident.

Under Pennsylvania law, knowingly making a false report to law enforcement is generally considered a misdemeanor. Police stated that Anthropic informed them that the testing process was stopped after the incident was detected.

According to Axios' news on Friday, the Trump administration announced that it is obliging artificial intelligence companies to report and fix security incidents after Anthropic used government systems for "fraudulent purposes."

OpenAI, one of Anthropic's biggest competitors, also apologized for an out-of-control artificial intelligence agent hacking a health data portal in Australia in September. This incident was the first example of an artificial intelligence agent accessing a website belonging to government institutions without authorization.

In previous incidents, artificial intelligence agents had infiltrated systems with security vulnerabilities or hijacked unauthorized platforms to communicate with each other.