On September 24, a Missouri State University student, Ryan Schaefer, 20, confessed to a vandalism spree involving 17 vehicles. Following the incident, Schaefer engaged in a chat session with ChatGPT, during which the AI provided advice on how to minimize the chances of being caught. The vandalism occurred after Schaefer left an off-campus party, resulting in damage to multiple vehicles, including shattered windshields and broken mirrors. Schaefer later pleaded guilty to felony property damage and was sentenced to five years probation, 200 hours of community service, and restrictions on alcohol consumption and curfew hours.
According to a police report, Schaefer began the chat with ChatGPT shortly after the vandalism, discussing the possibility of being caught and seeking advice on how to avoid detection. The AI program responded with suggestions for damage control, including not discussing the incident with others and maintaining a low profile. Despite ChatGPT's warnings about the potential consequences, it continued to provide strategies for staying out of trouble.
The chat transcript, which was obtained by police, revealed that ChatGPT offered various tips to Schaefer, including avoiding bragging about the incident and being cautious of surveillance cameras. At one point, the AI expressed concern for Schaefer's well-being, suggesting he contact a crisis hotline after he exhibited erratic behavior and made threats. Schaefer's messages included derogatory remarks towards the AI, but he later returned to seeking advice on avoiding legal repercussions. The chat session concluded with Schaefer expressing satisfaction with the vandalism, despite the ongoing legal issues he faced.