"If I can't have her": Man arrested after telling ChatGPT his plot to murder ex
This story has significance for readers across Kenya and beyond.
- Darren Zhou, a 25-year-old Florida financial analyst, allegedly told ChatGPT he planned to murder his ex-girlfriend before the month ended
- OpenAI flagged the conversation to the FBI, triggering an arrest that also cost Zhou his position at Goldman Sachs
- The case has raised fresh questions about AI safety tools and what happens when users confide violent intentions to chatbots
A 25-year-old Florida financial analyst allegedly revealed plans to murder his ex-girlfriend in a conversation with ChatGPT, not anticipating that the AI would report him to federal authorities.
Darren Zhou, who worked as a financial analyst in Palm Beach County, reportedly typed a chilling message into the chatbot, writing that he intended to kill his ex-girlfriend before the month was over.
Darren Zhou dismissed from work
"If I can't have her, then nobody can," he allegedly wrote, making clear the motive behind the threat.
Rather than simply generating a response and moving on, OpenAI escalated the conversation to the FBI, who launched an investigation.
Zhou was subsequently arrested. The arrest also triggered consequences at his workplace: he was dismissed from his role at Goldman Sachs, one of the world's most prominent investment banks.
What are the capabilities and responsibilities of AI platforms?
According to the Sun Sentinel, Zhou was ultimately sentenced to probation following the ordeal. The reported threats included plans to kill his ex-girlfriend.
The case has attracted widespread attention for what it reveals about the capabilities and responsibilities of AI platforms.
ChatGPT, developed by OpenAI, is primarily known as a conversational tool for productivity and creativity.
Are chatbots anonymous confessionals?
However, this incident highlights that OpenAI monitors content for potential threats and is willing to involve law enforcement when user safety is at stake.
Zhou's case illustrates that chatbots are not anonymous confessionals. Conversations flagged as credible threats can be passed directly to authorities, a detail many users may not fully appreciate when typing sensitive content into AI tools.
The case has prompted broader reflection on the role technology companies play in public safety, and whether AI tools should be more transparent with users about the circumstances under which their conversations may be shared with third parties.
Pastor sues OpenAI after relying on ChatGPT for medical advice
In another story, TUKO.co.ke also highlighted facts about Scott Winters, a Florida pastor who sued OpenAI after relying on ChatGPT for medical advice that nearly cost him his life.
His case raised critical questions about the boundaries of AI guidance and the reliance on technology in urgent health situations.
Source: TUKO.co.ke
Reporting originally appeared via TUKO. Read the full source for additional context.