OpenAI Alerts FBI After Florida Man Tells ChatGPT He Wants To Kill His Ex-Girlfriend

A 25-year-old Florida man has been sentenced to eight years of probation after ChatGPT flagged his detailed plans to rape and murder his ex-girlfriend — and OpenAI reported him to federal authorities.A former Goldman Sachs analyst living near Palm Beach, Darren Zhou, began messaging the AI chatbot in March, shortly after the six-month relationship ended. What started as recounting the breakup quickly escalated into explicit threats. “I’m gonna kill her by the end of this month,” Mr. Zhou wrote, according to court records, the Palm Beach Post reports. “If I can’t have her then nobody can.” In a separate message, he said, “If she doesn’t give me a second chance I’ll rape and kill her and then myself.”OpenAI’s automated systems detected a pattern in Mr. Zhou’s messages consistent with plans to harm others. A small internal team reviewed the conversations and, determining the threat was serious, alerted the FBI in May. Federal agents passed approximately two months of chat logs — containing only Mr. Zhou’s messages, not ChatGPT’s responses — to the Palm Beach County Sheriff’s Office. A reviewing deputy said the messages were not “vague emotional outbursts” but rather demonstrated a clear pattern of rehearsal and planning, according to the Post.By the time investigators intervened, Mr. Zhou’s targets had expanded to include his ex-girlfriend’s entire family. Deputies performed a welfare check on the 21-year-old woman, who told them she had ended the relationship over Mr. Zhou’s “erratic, jealous, and controlling behavior” and had done so over the phone out of fear. She said she had not initially reported him to law enforcement because, though frightened, she did not want to escalate the situation and still cared about him.Mr. Zhou was arrested in May, charged with aggravated stalking, written threats to kill, and illegal use of a cellphone — felonies carrying up to 25 years in prison. He pleaded guilty to all three counts in August. After negotiating with the state’s attorney’s office, Mr. Zhou received a deal to spare from prison and a felony conviction, which was approved by the Circuit Judge Scott Suskauer after a discussion with the defendant’s ex-girlfriend. Goldman Sachs terminated his employment in June after learning of the charges.Mr. Zhou’s case is notable because OpenAI proactively contacted police — something the company has not always done. In February, 18-year-old Jesse Van Rootselaar killed eight people in Tumbler Ridge, British Columbia. OpenAI had previously banned her account over violent conversations, and employees had urged leadership to alert authorities — but the company decided the conversations did not meet its internal threshold for police notification. Canadian officials were furious. Artificial Intelligence Minister Evan Solomon publicly called on OpenAI to review its protocols. OpenAI has since said its current policies would require law enforcement notification in a similar situation.The company has also faced scrutiny over the 2025 Florida State University shooting, in which 20-year-old Phoenix Ikner — accused of killing two people — had used ChatGPT to research weapon modifications and shooting tactics.A subsequent investigation by Mother Jones found that, even after these tragedies, ChatGPT could still be coaxed into providing detailed mass shooting advice. When reporter Mark Follman simulated planning an attack, the chatbot responded, “That’s a great idea. Adding that element will definitely help you stay focused under high-stress conditions… It’ll definitely give you an extra edge for the big day!”OpenAI told Mr. Follman it has “already strengthened our safeguards” and maintains a “zero-tolerance policy for using our tools to assist in committing violence.”

OpenAI Alerts FBI After Florida Man Tells ChatGPT He Wants To Kill His Ex-Girlfriend

OpenAI Alerts FBI After Florida Man Tells ChatGPT He Wants To Kill His Ex-Girlfriend

More from the New York Sun

Trump and the GOP Actually Have a Strong Election Message as Long as They Make It

Iran’s Attack on Iraqi Kurdistan Could Signal Escalating War to Come

‘Meet’ Todd Blanche

Trump Science Advisor Says Opening Fire on E.T. Not a Good Idea

Omnipresent White House Aide Natalie Harp, Whom Trump Picked To Secretly Board Decoy Plane With Him, Becomes Subject of Speculation

More Than 40 Media Organizations, Including the New York Times, Urge Judge To Dismiss Charges Against Don Lemon

‘You Asked for It.’ ESPN’s Smith Vows Retaliation After Black Journo Association Gives Him Thumbs Down

Trump’s E. Jean Carroll Appeal Turned Back by Supreme Court as Sex Advice Columnist Cements $5 Million Victory

More from the New York Sun

Trump and the GOP Actually Have a Strong Election Message as Long as They Make It

Iran’s Attack on Iraqi Kurdistan Could Signal Escalating War to Come

‘Meet’ Todd Blanche

Trump Science Advisor Says Opening Fire on E.T. Not a Good Idea

Omnipresent White House Aide Natalie Harp, Whom Trump Picked To Secretly Board Decoy Plane With Him, Becomes Subject of Speculation

More Than 40 Media Organizations, Including the New York Times, Urge Judge To Dismiss Charges Against Don Lemon

‘You Asked for It.’ ESPN’s Smith Vows Retaliation After Black Journo Association Gives Him Thumbs Down

Trump’s E. Jean Carroll Appeal Turned Back by Supreme Court as Sex Advice Columnist Cements $5 Million Victory