Prosecutors use ChatGPT logs as evidence in Palisades fire arson trial

Federal prosecutors trying a man accused of igniting one of the deadliest wildfires in Los Angeles history didn’t just rely on traditional evidence. They pulled his ChatGPT conversation logs.

Jonathan Rinderknecht, a 30-year-old dual French-US citizen and former Uber driver, faces charges of arson affecting interstate commerce and destruction of property by fire for allegedly starting the Lachman Fire near Pacific Palisades shortly after midnight on January 1, 2025. That fire, whipped by strong winds, escalated into the catastrophic Palisades Fire, which killed 12 people and destroyed more than 6,000 structures. If convicted, he faces up to 45 years in prison.

AI conversations as courtroom evidence

The prosecution’s case leaned on a mix of iPhone location data, security camera footage, and witness testimony. According to prosecutors, Rinderknecht used ChatGPT to generate images of fire, asked the chatbot “Why am I so angry all the time?”, and went on rants about how the wealthy were destroying the world. Perhaps most damning: a screen recording showed him asking ChatGPT whether someone could be blamed for a fire if it was lit by their cigarette.

The defense pushed back hard on the AI evidence, arguing that the ChatGPT interactions were taken out of context. Their broader position was that Rinderknecht was being used as a scapegoat. One juror noted during proceedings that they personally had similar types of conversations with AI chatbots, seemingly undermining the prosecution’s implication that the queries were inherently sinister.

The jury ultimately couldn’t reach a verdict. A mistrial was declared on June 26-27, 2026, with a retrial now scheduled for October 19, 2026.

The privacy problem nobody’s ready for

Every prompt you type into ChatGPT, Claude, Gemini, or any other AI assistant is stored somewhere. OpenAI’s privacy policy makes clear that conversation data is retained, and it can be produced in response to valid legal process. The Rinderknecht case is one of the first high-profile instances where AI chatbot logs have been introduced as substantive evidence in a federal criminal trial.

AI chat logs are qualitatively different from traditional search history. They’re conversational, they capture follow-up questions, and they can reveal a user’s reasoning process in a way that a list of search queries simply cannot. The difference between searching “can cigarettes start fires” and having a back-and-forth conversation with an AI about culpability for cigarette-caused fires is significant, both in terms of what it reveals and how a jury perceives it.

Defense attorneys in the Rinderknecht case argued that people say all kinds of things to chatbots — that curiosity isn’t intent, and venting isn’t planning. The hung jury suggests at least some jurors found that argument persuasive.

For users of centralized AI services, the practical takeaway is straightforward. OpenAI, Google, Anthropic, and others all have data retention policies and legal compliance obligations. Some already offer options to disable chat history, but the extent to which that actually prevents data from being recoverable in response to a subpoena varies by provider and jurisdiction.

The retrial in October will be worth watching closely. If prosecutors succeed the second time around, the Rinderknecht case will establish a precedent for using AI chat logs in criminal proceedings. If the defense prevails again, it may signal that juries aren’t yet ready to treat casual AI conversations as evidence of criminal intent.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.