OpenAI says it discovered and shut down ChatGPT accounts being used by a cyber-scam operation based in Poipet, Cambodia—a city notorious for scam centers run by Chinese criminal gangs. The tip came from WhatsApp, whose trust-and-safety team flagged unusual patterns in messages being sent through the platform. The scammers were using ChatGPT at every stage of their operation: creating convincing fake online personas, writing and translating messages sent to victims, generating fraudulent stock confirmations and fake legal notices, and producing recruitment fliers promising Indian nationals free flights, work visas, and accommodation—only to traffic them into forced labor in the scam compound.
The operation ran investment fraud, romance scams, gambling schemes, and law enforcement impersonation attacks simultaneously, often combining multiple types within a single campaign. OpenAI found that internal communications at the scam center also relied on ChatGPT: staff used it to draft announcements, track employee “debts” and disciplinary fines, and translate discussions about workers’ immigration status. Some conversations in the banned accounts referenced detention and escape attempts by trafficked workers. OpenAI estimates the operation “may have interacted with hundreds of targets across multiple scam types,” with some victims losing thousands of dollars. This disclosure adds to a growing body of evidence that AI tools have become a core component of organized online fraud: INTERPOL published research the same day finding that AI is now used to automate more than half of all cybercrime activity across Africa.