OpenAI played a key role in the arrest of a Florida man who turned to ChatGPT to plot a violent revenge scheme against his ex-girlfriend, according to court documents and local reporting.
Darren Zhou, 25, a former Goldman Sachs analyst from South Palm Beach, Florida, was arrested in May after authorities say he used the AI chatbot to detail plans to stalk, kidnap, rape, and murder a 21-year-old woman he had dated for six months.

Zhou’s messages to ChatGPT included statements such as “I’m gonna kill her by the end of this month” and “If I can’t have her then nobody can,” according to court records cited in the Daily Wire report. He also described waiting for her at her gym with weapons, planning to force her to his home, and sending her photos of zip ties, latex gloves, and firearms.
OpenAI’s safety protocols flagged the explicit threats of violence, and the company handed over two months’ worth of Zhou’s chat logs to federal agents, who passed them to the Palm Beach County Sheriff’s Office.

Chat logs showed ‘pattern of rehearsal and planning’
The Palm Beach Post reported that a deputy who reviewed the logs—which contained only Zhou’s messages, not ChatGPT’s responses—said they were not “vague emotional outbursts” but instead showed a pattern of rehearsal and planning.
While Zhou was messaging the AI, he was also allegedly harassing his ex in real life. She had blocked him on multiple platforms due to his stalking behavior, but he used burner apps to send sexually explicit texts and chilling location-based messages, such as sending just the name of her gym while she was inside.

Zhou, a Northeastern University graduate, faced felony charges including aggravated stalking, written threats to kill, and illegal use of a cell phone. The charges carried a potential sentence of up to 25 years in prison.
Plea deal spares prison time
In a surprising courtroom twist, Zhou avoided prison and a felony conviction. State prosecutors struck a plea deal on August 13 that allowed Circuit Judge Scott Suskauer to withhold judgment, instead sentencing Zhou to eight years of probation.
Judge Suskauer noted he only approved the deal because the victim agreed to it. “Get over it and move on,” he told Zhou. “No one deserves the type of conduct that you exposed this lady to.”
Zhou’s attorney described the episode as a “difficult mental health episode,” downplaying the seriousness of the threats and claiming Zhou did not actually own guns and was “not a violent person.”
As part of his sentence, Zhou will wear an ankle monitor for two years and complete court-mandated counseling.
Source: www.dailywire.com — https://www.dailywire.com/news/chatgpt-reports-man-to-fbi-after-disturbing-threats
