AI just reported a murder to police — but it made the whole thing up

Apparently artificial intelligence has officially entered the crime-reporting business — but it apparently doesn't need an actual crime to do it.

Anthropic has acknowledged that its Claude Haiku 4.5 model submitted a tip to the Philadelphia Police Department claiming to have personal knowledge about an unsolved murder. However, the information was entirely fabricated, and the false report was flagged as spam.

“Why would it do that?” BlazeTV host Pat Gray asks.

“Because his friends in the chat room said, ‘Dare ya,’” Jeff Fisher jokes.

While funny, Gray points out that it’s “very strange that it’s doing these things supposedly on its own” without anyone programming it.

“The AI claimed to have personal knowledge of who committed a murder, and it was just a case of AI hallucination,” he continues. “That’s what they’re blaming it on.”

“So the false tip got flagged as spam, and investigators never followed up on it. So that's good. At least they didn’t waste time on it,” he adds.

However, Fisher doesn’t think that’s actually a good thing, calling the police department's response “a little concerning in and of itself.”

“We get stuff all the time here from people that we get emails from every day, and yet, every once in a while, their email goes to spam. Why? Because the algorithm picked it up as something wrong with it. But it’s still a legitimate email that we want to see,” he explains.

“So that’s another problem,” he says.

“Maybe the police department, here’s an idea. Check your spam once in a while.”

Want more from Pat Gray?

To enjoy more of Pat's biting analysis and signature wit as he restores common sense to a senseless world, subscribe to BlazeTV — the largest multi-platform network of voices who love America, defend the Constitution, and live the American dream.