A 25-year-old former Goldman Sachs analyst, Darren Zhou, pleaded guilty on August 13 to felony stalking and threat charges after OpenAI flagged his ChatGPT conversations outlining plans to harm his ex-girlfriend and alerted the FBI, according to medianama.com. Zhou received eight years of probation and two years of electronic monitoring under a plea deal that avoided prison time and a felony conviction.
Zhou was charged in June with aggravated stalking, written threats to kill, and illegal use of a cell phone, felonies that carried a combined maximum sentence of 25 years. He pleaded guilty to all three counts, and Circuit Judge Scott Suskauer accepted the plea because Zhou’s ex-girlfriend approved the agreement. Zhou spent two days in jail after his May arrest before posting $100,000 bail and was later terminated from Goldman Sachs. OpenAI’s internal team detected a pattern in his ChatGPT conversations and forwarded the tip to law enforcement.
OpenAI’s policy involves reviewing ChatGPT conversations when the system detects potential harm plans, forwarding serious cases to authorities. This case highlights the role of AI companies in monitoring user activity for public safety. Zhou’s case is a rare example of AI-generated content leading directly to law enforcement intervention, underscoring the increasing intersection of AI technology and criminal justice.
Zhou’s probation and electronic monitoring will last a combined ten years, with the felony record withheld if he completes probation successfully, the court confirmed. The case marks one of the first instances where OpenAI’s content monitoring directly contributed to preventing a violent crime, as reported by medianama.com.