Disruption of a Criminal Network
On July 31, 2026, OpenAI announced the disruption of a Cambodia-based criminal network that utilized ChatGPT to facilitate various fraudulent schemes. The operation leveraged the AI model to create fake personas, generate and translate messages, produce promotional materials, and fabricate documents used against victims.
Investigation and Findings
OpenAI’s investigation traced the coordinated network to accounts likely originating in Cambodia, specifically around Poipet in Banteay Meanchey province. The inquiry was initiated following a lead from WhatsApp, leading to the sharing of threat signals with industry partners and relevant authorities.
Scam Tactics Employed
The network operated multiple types of scams simultaneously, often blending tactics. Operators used dating personas to establish trust before introducing fraudulent investment opportunities, including schemes related to cryptocurrencies and spot gold trading. Other accounts impersonated gambling platforms offering fake bonuses or posed as law enforcement to demand payments for alleged offenses.
Patterns of Deceptive Behavior
OpenAI identified a recurring pattern in the network’s interactions with targets, characterized by three stages: outreach (the “ping”), emotionally charged messages (the “zing”), and requests for money or personal information (the “sting”). Scammers also generated forged documents, including passports, legal notices, stock-purchase confirmations, and fake trading interfaces.
Indicators of Human Trafficking
In addition to scams, some content produced by accounts in the network suggested potential links to human trafficking and forced labor. This included social media job advertisements for “chatter” roles in Poipet, promising flights, visas, and work permits. Records of employee debts, salary deductions, and discussions about immigration status were also noted. Conversations referenced detention and escape attempts, although OpenAI clarified that these exchanges do not confirm the circumstances of specific individuals.
Response and Impact
OpenAI banned the ChatGPT accounts associated with this operation and shared relevant indicators with partners and authorities to prevent these actors from regaining access. The operation reportedly interacted with hundreds of targets, with conversations indicating that victims may have lost thousands of dollars, although OpenAI could not independently verify these claims.
Original source: OpenAI News