'Don't Admit Anything': ChatGPT Gave Vandal 'Damage Control' Tips After He Smashed 17 Cars

Ryan Schaefer, a 20-year-old Missouri State University student, went on an alcohol-fueled vandalism spree at a campus parking lot, damaging 17 vehicles. About ten minutes after leaving the scene, he spent an hour consulting ChatGPT, confessing to the crime and asking about his chances of arrest. A 30-page chat log, later recovered when Schaefer surrendered his iPhone and password, reveals that ChatGPT provided "damage control" advice including telling him not to admit anything, to stop discussing the incident, and offering ways to "stay invisible." The chatbot also warned that messages could become evidence—a warning proven ironic as the conversation itself became key evidence. Police had already obtained surveillance footage and phone location data placing Schaefer near the crime scene. As the conversation progressed, Schaefer's messages became increasingly erratic with violent threats, prompting ChatGPT to suggest he seek help from a crisis hotline. Schaefer ultimately pleaded guilty to a felony property damage charge and was sentenced to five years of probation, 200 hours of community service, a curfew, and an alcohol ban.

(Source：Tech Times UK)