Prosecutors used ChatGPT logs in LA wildfire arson case that ended in mistrial

In short: Prosecutors in a federal arson case tied to the 2025 Palisades Fire used ChatGPT logs as part of their evidence, but the first trial ended with no verdict.

What happened

Jonathan Rinderknecht, 29, is accused of starting a brush fire near Pacific Palisades on January 1, 2025. Investigators say that fire, sometimes called the Lachman Fire, may have smoldered underground and then flared back up days later as the Palisades Fire during strong Santa Ana winds. The Palisades Fire has been described as one of the most destructive wildfires in Los Angeles history.

Rinderknecht was indicted in federal court in October 2025. He has pleaded not guilty. He faces three federal arson related charges and prosecutors say he could face up to 45 years in prison if convicted on all counts.

At trial, prosecutors pointed to several types of evidence, including phone location data that they say placed him near where the fire started. They also cited witness statements, a lighter found near the origin area, and his behavior after the fire, including following firefighters and filming the scene.

Prosecutors also used records of his ChatGPT activity. Reporting says this included questions like “Why am I so angry all the time?” and a question about whether someone would be responsible for a fire started by a cigarette. Investigators also said he asked ChatGPT how to delete video messages from iCloud (Apple’s online storage, like a remote hard drive).

The judge blocked some AI generated image evidence from being shown to the jury. The first trial ended in a mistrial after jurors could not reach a unanimous decision.

Why it matters

This case shows how chats with an AI tool can end up treated like other digital records in court, similar to text messages or search history. It also shows judges may limit AI created content if they think it could unfairly sway a jury.

Source: The Verge AI