ChatGPT logs backfire as evidence in Palisades fire arson trial, ending in mistrial

Prosecutors in the arson trial of Jonathan Rinderknecht, who faced charges related to a devastating New Year's Day 2025 wildfire, built their case using an unusual combination of evidence: iPhone location tracking, security camera recordings, witness accounts—and records of his conversations with ChatGPT. Yet the strategy failed, culminating in a deadlocked jury and a mistrial.

What prosecutors presented#

The prosecution highlighted several elements from Rinderknecht's ChatGPT interactions. They noted that he had used the AI tool to generate images of fire, posed existential questions to the chatbot like "Why am I so angry all the time?", and vented about wealth inequality and environmental destruction. They also introduced a screen recording in which he asked the AI whether someone could be held accountable for a fire started by their cigarette.

The jury's verdict#

Despite this digital evidence, the jury remained unconvinced of guilt. The panel deadlocked 10-2 in favor of acquittal, prompting the judge to declare a hung jury and declare a mistrial.

A juror who spoke with CBS LA explained the disconnect, stating that she dismissed the ChatGPT records as meaningful proof. "I talk to ChatGPT all the time," she said, adding that she found it objectionable for prosecutors to frame ordinary chatbot conversations as a window into criminal intent or character flaws.

The larger question#

The case underscores broader questions about how evidence derived from AI conversations should be evaluated in the courtroom—and whether typical interactions with generative AI tools can reliably signal guilt or motive.