ChatGPT logs were at the center of a historic arson trial in the US, marking a new era in the use of artificial intelligence as evidence. Defendant Jonathan Rinderknecht faced arson charges for the New Year's Eve 2025 fire , which became one of the deadliest fires in Los Angeles history . The trial ended in a mistrial due to a deadlocked jury, but it left a crucial legal precedent behind it.
See also: ChatGPT Images 2.0: New version of OpenAI's AI image generator
According to prosecutors, Rinderknecht intentionally lit the so-called Lachman Fire on January 1, 2025.The fire burned underground for a week before rekindling on January 7, 2025 , in high winds, turning into the devastating Palisades Fire. To prove the defendant's intent, prosecutors used location data from iPhone , security camera footage, witness statements — and ChatGPT logs .
Chat records revealed that the defendant had asked ChatGPT to create images of a burning forest and crowds running for safety months before the fire, and had also created images depicting fires and wealth inequality. He also asked the chatbot why it was so angry all the time and expressed frustration with the way the rich were destroying the world. In a videotaped screengrab, he asked ChatGPT if someone could be held responsible for a fire that his cigarette started.
ChatGPT logs: A new kind of digital evidence
This case is a landmark for the legal community, as for the first time ChatGPT logs have been used as direct evidence of intent and motive in a criminal trial. Legal analysts point out that AI can now bridge the gap between thought and action in criminal law. Chatbot conversation logs are increasingly being treated as digital fingerprints, similar to emails or social media history.
The defense argued that there were no internet searches related to arson or incendiary materials on the defendant’s devices. Prosecutors countered that the defense opened the door with this argument, allowing the ChatGPT logs as evidence revealing the defendant’s frustration with his financial situation and social isolation. Also important was Rinderknecht to delete video messages from iCloud immediately after the search warrant was executed — a fact that prosecutors interpreted as evidence of guilt.
See also: How to write better prompts in ChatGPT?
Mistrial: Why the ChatGPT logs didn't convince the jury
Despite the seriousness of the charges — Rinderknecht faces up to 45 years in prison if convicted — the jury was not convinced. The trial ended in a deadlock with a 10-2 vote in favor of the defense, forcing the judge to declare a hung jury and a mistrial. One juror, Syrina, told CBS LA that she didn’t believe the ChatGPT logs proved anything, saying she talks to ChatGPT all the time. She even expressed anger that using the chatbot was portrayed as a sign of moral weakness.
This backlash highlights a critical reality: ChatGPT has over 100 million weekly active users worldwide, and conversations with AI chatbots have become part of the daily lives of millions of people. This makes it difficult to legally assess them as evidence of intent, as people express thoughts, feelings, and fantasies to chatbots without intending to act.
The case has significant implications for OpenAI and other AI, such as Google with Gemini, Anthropic with Claude , and Microsoft with Copilot. These companies may face increased scrutiny over their data retention policies and access by law enforcement. Users should be aware that records of conversations with AI are not private and can be seized with a warrant, revealing intent, motivation, or guilt.
See also: OpenAI adds Lockdown Mode to ChatGPT to prevent data theft
A mistrial does not mean an acquittal — prosecutors can retry the case. Meanwhile, victims of the Palisades Fire can pursue civil lawsuits for damages, without the constraints of a criminal conviction. The case is a landmark for the use of AI in the justice system and is expected to influence future trials, AI data privacy , and law enforcement digital search protocols.
