OpenAI has dismantled a coordinated scam operation based in Cambodia that used ChatGPT to support a wide range of online fraud, including investment scams, romance scams, gambling schemes, and fake law enforcement impersonation campaigns.
According to the company, the network likely operated from Poipet, a city in western Cambodia that has previously been linked to large-scale scam compounds and human trafficking. OpenAI banned the accounts after discovering they were using its AI models to create fake online identities, generate and translate messages, produce promotional materials, and assist with the day-to-day management of the criminal operation.
The scammers used ChatGPT to write convincing messages, build fake social media profiles, and create advertisements recruiting workers for so-called “chatter” jobs in Poipet. The ads targeted people in Bangladesh and India by promising salaries of around $800 per month, bonuses, free flights, accommodation, meals, one-year Cambodian visas, and work permits.
OpenAI also found that some accounts relied on ChatGPT for internal administrative tasks, including drafting staff announcements, translating communications, and maintaining records related to employee debts, salary deductions, fines, loan repayments, visa overstays, immigration status, work permits, and recruitment incentives.
The investigation was carried out in collaboration with WhatsApp. OpenAI said the case highlights how modern cybercriminal organizations no longer focus on a single type of fraud. Instead, they rapidly switch between different scams and use whatever tactics they believe are most likely to deceive victims.
The network ran several fraud campaigns at the same time. Some operators posed as romantic partners on dating platforms before convincing victims to invest in fake cryptocurrency or spot gold trading schemes. Others pretended to represent online gambling platforms, promising non-existent bonuses and winnings to lure users into making deposits. Another group impersonated police or government authorities, falsely accusing victims of serious crimes and demanding payments to avoid legal consequences.
To make the scams more convincing, the criminals generated forged documents and fake digital assets, including passports, legal notices, stock purchase confirmations, gambling platform interfaces, and payment screenshots.
OpenAI said the scammers followed what it describes as a “ping-zing-sting” strategy. Victims were first contacted through messaging platforms such as WhatsApp and Telegram, then gradually drawn into long conversations designed to build trust. Once the relationship was established, victims were instructed to send deposits, activation fees, or bogus fines and were shown fabricated proof of successful transactions to encourage further payments.
The company also uncovered evidence suggesting some of the activity was connected to human trafficking and forced labor. According to OpenAI, organized crime groups in East Asia often recruit workers with promises of high-paying jobs, only to confiscate their passports and force them to work in scam compounds under abusive conditions.
If this article helped you, please consider supporting our work. Every small contribution keeps Abijita.com independent and running.
OpenAI said the full financial impact of the operation remains unknown. However, internal communications reviewed during the investigation indicate the network may have targeted hundreds of victims across multiple fraud campaigns, with some conversations referring to individual losses worth thousands of dollars. The company noted it could not independently verify those claims.
