Your ChatGPT logs are now court evidence — and they just caused a mistrial

Federal prosecutors in Los Angeles attempted to convict Jonathan Rinderknecht, a former Uber driver, of starting the devastating Palisades wildfire that destroyed thousands of homes. The cornerstone of their case wasn't DNA or eyewitnesses, but rather thousands of deeply personal conversations the defendant had with ChatGPT over several months.

The digital diary revealed some incredibly bizarre interactions. The prosecution showed the jury how Rinderknecht asked the AI to generate images of rich people ignoring a raging fire, repeatedly asked "Why am I always angry?", and even screenshotted a query asking if a person is legally responsible if a wildfire accidentally starts from a discarded cigarette. Prosecutors framed this last question as a calculated move to pre-engineer an alibi.

The defense team argued these queries were taken entirely out of context. Surprisingly, the strategy to paint these chats as the digital manifestations of a supervillain backfired when one juror admitted they also have weird, deeply personal conversations with AI, immediately normalizing the defendant's bizarre chat history.

Because US federal criminal trials require a unanimous verdict, the jury's 10-to-2 split in favor of acquittal forced Judge Ann Hwang to declare a mistrial. The government, however, is not backing down and has already scheduled a retrial for October 19.

Using conversational AI as an emotional dumpster fire is no longer a private hobby. The upcoming trial will set a chilling precedent for whether digital therapy sessions with a corporate server can legally prove criminal intent. It seems the "delete chat" button is nothing more than a placebo for peace of mind.

Source: The Verge