A court in Palm Beach County, in southern Florida, sentenced Darren Zhou, a 25-year-old former financial analyst, to eight years of probation after he used ChatGPT to plan the kidnapping, sexual assault, and crime against his ex-girlfriend, in a case that resulted in an emergency alert from OpenAI directly to the FBI.
Zhou was working as an analyst at Goldman Sachs in West Palm Beach when, in March 2026, weeks after his girlfriend of six months ended their relationship, he began sending messages to the chatbot with detailed plans for violence.
According to the , the conversations escalated quickly from accounts of the breakup to explicit threats.
"I'm going to kill her before the end of this month," Zhou wrote to the chatbot in one of the messages preserved in the files. In another, he added, "If I can't have her, then nobody can."
The plans he described to the artificial intelligence included waiting for her in the parking lot of her gym with flowers; if she rejected him, he would "pull out a weapon, kill her, and then shoot himself."
He also sent the victim photographs of an AR-15 rifle, a Glock pistol, a 12-gauge shotgun, plastic ties, and latex gloves.
Meanwhile, Zhou harassed the young woman from various numbers generated by applications, ignoring her repeated blocks.
On her 21st birthday, he sent her messages with explicit threats of sexual assault, and one of the texts contained only the name of the place where he was at that moment.
The victim acknowledged to investigators that she had not reported Zhou because, despite her fear, she did not want his behavior to escalate and still felt affection for him.
OpenAI issued an emergency alert to the FBI, which transferred two months of conversation records to the Palm Beach County Sheriff's Department.
The detectives cross-referenced those records with the anonymous messages received by the young woman and concluded that Zhou represented "a credible threat."
He was arrested in May, spent two days in jail, and was released after paying a bail of 100,000 dollars. Goldman Sachs terminated his employment in June upon learning of the charges.
The prosecutor accused him of aggravated harassment, written death threats, and illegal use of a communications device, offenses that could carry up to 25 years in prison.
However, on Wednesday Zhou pleaded guilty to all three charges and reached a deal that allowed him to avoid both prison and a formal felony conviction.
The judge Scott Suskauer accepted the agreement because the victim herself approved it. Before issuing the sentence, he addressed the accused with some direct words: "Get over it and move on. There are other people in the world. No one deserves the kind of behavior you subjected this woman to."
Defending attorney Steven Bell attributed his client’s behavior to "a very difficult mental health episode" and highlighted that Zhou graduated magna cum laude from Northeastern University with no prior criminal record.
The case adds to a series of incidents in which ChatGPT has been linked to plans for violence.
The Attorney General of Florida opened an investigation against OpenAI in April due to the possible connection of the chatbot with the shooting at Florida State University in 2025, perpetrated by someone who had had over 270 conversations with the tool in the previous year.
In June, OpenAI also alerted the FBI about a man in Brazil who planned to kill his eight-year-old son to evade paying child support, which allowed for his timely arrest.
As part of his sentence, Zhou must wear an ankle monitor for the first two years, complete a program for offenders, undergo a mental health evaluation, and refrain from any contact with the victim, as well as from possessing weapons or consuming drugs and alcohol.
Related videos:
Filed under: