More than eight months after the devastating Palisades Fire in Los Angeles County was finally contained, the Department of Justice on Wednesday announced that they had made an arrest in connection with the massive January blaze, which killed 12 people and destroyed thousands of structures in coastal communities, becoming one of the most destructive wildfires in California history.
Jonathan Rinderknecht, 29, was apprehended Tuesday near his residence in Melbourne, Florida, and has been federally charged with “destruction of property by means of fire,” acting U.S. Attorney Bill Essayli said at a press conference on Wednesday. This felony charge, he added, carried a mandatory minimum prison sentence of five years in federal prison but is punishable by up to 20 years in prison. Rinderknecht was expected to make his first appearance in a federal court in Orlando later that day, Essayli said. Rinderknecht has not yet entered a plea.
But more curious than the allegation that a Florida man was responsible for setting a small brush fire on the other side of the country that ultimately ravaged the entire region was the “digital evidence” in the criminal affidavit that Essayli presented to the media. Investigators, he noted, allege that some months prior to the burning of the Pacific Palisades, Rinderknecht had prompted ChatGPT to generate “a dystopian painting showing, in part, a burning forest and a crowd fleeing from it.” A screen at the press conference showed several iterations on such a concept, some including buildings engulfed in flames.
It’s not clear whether this is the first time a user’s ChatGPT history has been used as evidence against them in a criminal case, nor whether OpenAI, the tech company that developed it, collaborated with law enforcement in their investigation of Rinderknecht. The AI firm’s policies state that they require “a subpoena, court order, search warrant or equivalent before disclosing requested non-content user information,” and only shares content “in response to a valid warrant or equivalent.” They also provide “only the data specified in the legal process,” the policy states. Sam Altman, OpenAI’s CEO, has warned that ChatGPT offers no legal confidentiality. And the AI for Change Foundation, a nonprofit that advocates for ethical AI with enhanced safeguards, has pointed out that court precedents are evolving in this area, with judges in the U.S. and abroad ruling that AI chatbot logs are discoverable records.
OpenAI did not immediately return a request for comment.
Essayli said that in addition to the ChatGPT pictures, the Justice Department’s allegations against Rinderknecht are supported by evidence found on his phone, alleged false statements to authorities, as well as “his behavior during the Lachman Fire,” which he is charged with “maliciously starting” on a hilltop in the Pacific Palisades shortly after midnight on New Year’s Day. The Lachman Fire was suppressed by firefighters — according to the criminal complaint, Rinderknecht called 911 several times to report it and then returned to take video of emergency responders — but continued to smolder in underground root systems until extreme Santa Ana winds, hot and dry, caused it to surface and spread on Jan. 7, then explode as the Palisades Fire. The same day, the Eaton Fire ignited in the San Gabriel Mountains of east L.A. County. That blaze ultimately killed 19 people.
While the official statement from the U.S. Attorney’s Office on the case does not mention the ChatGPT images discovered in Rinderknecht’s record of exchanges with the chatbot, Essayli seemed to suggest that they could be used in court as proof of premeditated intent to commit arson. He declined to address the question of possible motive. Rinderknecht had been working as an Uber driver in the area shortly before the Lachman Fire ignited, with two of his passengers from late that evening telling investigators that he appeared angry and agitated during their rides.
The legal matter of admitting ChatGPT evidence remains mostly uncharted. It’s likely, however, that attorneys and law enforcement will increasingly regard chatbot logs as forensically relevant to prosecutions in the same way text messages and online communications are. They can provide revealing insights into states of mind and individual preoccupations — fire, for example. One might imagine their secrets safer with an AI program than another person, but, as with all digital information, the data can always be uncovered.