Artificial Intelligence

ChatGPT Tips Off FBI, Leads to Goldman Sachs Analyst's Arrest

Is this a lifesaving breakthrough or a dangerous overreach by a private company with no accountability?

Published AUG 18

ChatGPT Tips Off FBI, Leads to Goldman Sachs Analyst's Arrest

Story

Above: Stock image of the Federal Bureau of Investigation seal. Image credit: Beata Zawrzel/NurPhoto/Getty Images

The Facts

Darren Zhou, 25, the then Goldman Sachs analyst who used ChatGPT to outline plans to harm his ex-girlfriend, has been arrested after OpenAI detected his threatening messages and alerted the FBI, which then passed two months of chat logs to the Palm Beach County Sheriff's Office.
USA Today
Medianama
Court records show Zhou wrote "I'm gonna kill her by the end of this month" and described waiting at his ex-girlfriend's gym armed with weapons, planning to force her to his home at gunpoint. He also claimed to have sent her photos of firearms, zip ties and latex gloves.
Daily Wire
India Today
A sheriff's deputy who reviewed the chat logs — which contained only Zhou's messages, not ChatGPT's responses — said they were not vague emotional outbursts but demonstrated a clear pattern of rehearsal and planning.
Mass. Lawyers Weekly
Medianama
Zhou was arrested in May and charged in June with aggravated stalking, written threats to kill and illegal use of a cell phone — felonies carrying a combined maximum sentence of 25 years in prison. He spent two days in Palm Beach County jail before posting $100,000 bail.
USA Today
ChosunBiz
Last week, Zhou pleaded guilty to all three charges under a plea deal negotiated with Assistant State Attorney Ana Cuskova. Circuit Judge Scott Suskauer withheld formal adjudication and sentenced Zhou to eight years of probation, saying he accepted the deal because the victim had approved it.
Mass. Lawyers Weekly
Medianama
The case drew comparisons to an earlier incident in which OpenAI did not alert Canadian authorities before a ChatGPT user killed eight people in Tumbler Ridge, British Columbia. OpenAI has since stated its current protocols would require police notification in a similar situation.
ChosunBiz
Medianama

Sources Split

LEFT

CENTER

RIGHT

ESTAB
CRIT

CENTER

PRO
ESTAB

The Spin

Narrative A

ChatGPT did what no one else would — it went to the FBI when a Goldman Sachs analyst spent months detailing his plans to kidnap, rape and kill his ex-girlfriend. The victim was too scared to report him herself, so an AI flagged the threat and handed it to federal law enforcement. This is a landmark moment showing AI firms can actively protect people when human systems fall short.

IT Guy on X

Vaibhav Sisinty on X

Narrative B

The ChatGPT-to-FBI pipeline worked here, but that doesn't mean the rules are clear or safe. OpenAI's internal team makes judgment calls about which conversations get handed to federal law enforcement — and no public rulebook explains where the line is. Trusting a private company to decide what's reportable, with no transparency, is a serious risk that this case doesn't resolve.

Surendra Koutarapu on X

Narrative C

Two disturbing cases this week raise urgent questions about AI safety. In Florida, OpenAI flagged months of ChatGPT conversations detailing plans to harm an ex-girlfriend and alerted the FBI. In Massachusetts, a 17-year-old faces murder charges after allegedly using ChatGPT to explore fantasies about killing his family. One case triggered intervention; the other didn't. The industry still lacks clear answers on what AI should do when users signal danger.