OpenAI Bans ChatGPT Accounts in Cambodia-Based Multi-Scam Operation
OpenAI banned a network of ChatGPT accounts used by a Cambodia-based scam operation. The accounts very likely originated in Cambodia and likely operated in or around Poipet.
OpenAI said the operation may have interacted with hundreds of targets through investment, romance, gambling and law-enforcement impersonation schemes. Total financial losses remain unknown.
The investigation began after a lead from OpenAI’s peers at WhatsApp.
How one network mixed scam formats
The network used ChatGPT to create fake personas, generate and translate messages, produce promotions and handle routine administration. Dating identities built trust before introducing fraudulent cryptocurrency investments and spot gold trading.
Others maintained romantic conversations, offered fake gambling bonuses and winnings, or impersonated law enforcement to demand fines. Users also generated images of forged passports, legal notices and trading or gambling interfaces.
The ping, zing and sting
OpenAI called the sequence the ping, zing and sting: outreach through WhatsApp, Telegram or social media; emotional pressure using secrecy, urgency, romance or guaranteed returns; then demands for deposits, fees or fictitious fines.
Human trafficking and forced-labor indicators
Some activity suggested human trafficking or forced labor. Users advertised “chatter” jobs in Poipet with flights, accommodation, meals, visas and work permits, while records tracked debts, salary deductions, fines and loan repayments.
Conversations also mentioned immigration status, detention and escape attempts. OpenAI could not establish individual circumstances, but said the material was consistent with public reporting on organized crime in Southeast Asia.
Impact and business relevance
The scammers’ own messages referred to individual victims losing thousands of dollars, but OpenAI could not independently verify those claims. It shared indicators with partners and authorities, and acted to impede renewed access.
Businesses should not assume different scam narratives represent separate risks. OpenAI said effective disruption must address both victim-facing fraud and the organizations that orchestrate and profit from it.