A man who has been arrested on suspicion of starting the deadly Palisades Fire that destroyed thousands of homes and killed a dozen people in Los Angeles last year allegedly used ChatGPT to generate images of burning forests and cities, according to law enforcement officials.
The suspect in custody, 29-year-old Jonathan Rinderknecht, was arrested near his home in Florida on Tuesday for destruction of property by means of fire, acting US Attorney Bill Essayli told reporters at a press conference on Wednesday.
Federal prosecutors allege that Rinderknecht started an initial eight acre blaze, called the Lachman fire, in the early hours of New Year’s Day. Firefighters initially put out the fire, but it rekindled days later, which prosecutors say eventually grew into the even larger inferno that devoured the Pacific Palisades hillside, causing numerous deaths and hundreds of billions of dollars in damages.
That night before the New Year, Rinderknecht had been working as an Uber driver, prosecutors say in a complaint. Two of his passengers told investigators that he seemed angry and agitated. After dropping off his last customer in the Palisades, Rinderknecht drove to the Skull Rock trailhead, and then walked to a clearing on the hilltop that a friend said he frequently visited. Cell data confirmed his location, the complaint said. While there, he took videos with his iPhone and listened to a French rap song called “Un Zder, Un The,” per the LA Times.
Soon, Rinderknecht turned to an AI chatbot for advice. Shortly after starting the fire after midnight, prosecutors allege, he tried calling 911 several times because of bad reception to report the fire. When he finally got through to an operator, he typed a question into ChatGPT, OpenAI’s chatbot.
“Are you at fault if a fire is [lit] because of your cigarettes,” Rinderknecht allegedly asked the AI model, per an affidavit quoted by the Los Angeles Times.
ChatGPT’s response: “Yes.”
After running from the fire, Rinderknecht claimed he returned to offer firefighters help with the blaze. But investigators argue that he had an ulterior motive. They allegedly found videos on his phone he had taken of firefighters trying to put out the fire, and a screen recording of his phone as he tried to call 911.
The videos and his ChatGPT question suggest “he wanted to create evidence regarding a more innocent explanation for the cause of the fire,” the complaint said, per the BBC.
The investigation also dug into Rinderknecht’s past interactions with ChatGPT. In July 2024, he asked the chatbot to generate an image of a “dystopian painting” that showed a crowd of people running away from a burning forest and trying “to get past a gigantic gate with a big dollar sign on it.”
“On the other side of the gate and the entire wall is a conglomerate of the richest people,” the prompt continues. “They are chilling, watching the world burn down, and watching the people struggle. They are laughing, enjoying themselves, and dancing.”
The resulting images vary, but all of the ones shared in the complaint show either a forest or a city engulfed in flames. One month before the fire, Rinderknecht also confided in ChatGPT that he felt “amazing” and “liberated” after burning a Bible, per the BBC’s coverage.
“You could see some of his thought process in the months leading up, where he was generating some really concerning images up on ChatGPT,” Essayli said at the conference, per ABC News.
Essayli, however, avoided answering questions about Rinderknecht’s possible motives. He also strongly emphasized that the the rap track the suspect was listening to had a music video featuring several objects like a metal trash barrel set on fire — which feels like a stretch, not to mention sounding like the age-old scapegoat of blaming rap music for crime. It also doesn’t appear that Rinderknecht’s prompts asked ChatGPT to depict a burning city; they only described a “burning forest.” The AI, it seems, generated scenes of burning cities on its own.
But if the allegations are true, this wouldn’t be the only time that AI chatbots have been used in deadly crimes. The Green Beret soldier who blew himself up in a Cybertruck outside the Trump Towers hotel in Las Vegas at the beginning of the year — the same day that Rinderknecht allegedly started the fire, in fact — asked ChatGPT to help plan his attack.
The news also comes during mounting concern over reports of “AI psychosis,” disturbing mental episodes in which a person who’s been interacting extensively with an AI chatbot develops severe delusions and suffers breaks with reality. Some of these episodes have ended in trips to the hospital, suicide, and murder, including a man who allegedly killed his mother after ChatGPT strengthened his conviction that she was part of a conspiracy to surveil him.
At this time, just how extensive Rinderknecht’s alleged interactions with the OpenAI product remain unclear — as are his motives.
More on ChatGPT: 13-Year-Old Arrested for Asking ChatGPT How to Kill His Friend