IN A NUTSHELL: Users have long been warned to be careful what they tell ChatGPT, especially if it could get them in trouble with the law. California prosecutors, for example, used the chatbot's logs in their case against the man accused of starting a blaze that became the Pacific Palisades fire of 2025.
Jonathan Rinderknecht was accused of deliberately starting last year's fire, which killed a total of 31 people and destroyed more than 6,000 buildings.
The prosecution's case included everything from his iPhone's location data and security camera footage to witness testimonies.
Prosecutors also used Rinderknecht's ChatGPT logs. They said he used the tool to generate an image depicting a city on fire, asked the AI "why am I so angry all the time?" and went on tirades about how the rich were destroying the world -- allegedly influencing his decision to start the blaze in a wealthy neighborhood.
Rinderknecht is said to have used ChatGPT like a diary. One prompt asked if he could be held responsible for a fire that was caused by his fallen cigarette.
