One network, several scam scripts
OpenAI said on July 31 that it banned a coordinated ChatGPT account network likely originating in Cambodia and operating around Poipet. The investigation began with a lead from WhatsApp and found investment, romance, gambling and law-enforcement impersonation schemes running in parallel; indicators were shared with industry and authorities.
According to OpenAI, accounts created dating personas and fake investment experts, translated victim messages, and generated images of passports, legal notices, stock confirmations and gambling interfaces. Conversations moved from trust building to crypto or spot-gold pitches, fictitious fines, activation fees and proof-of-payment requests.
Possible trafficking indicators
Some activity referred to job ads, worker debt, salary deductions, immigration status and escape attempts, which may indicate forced labor or trafficking. OpenAI cautions that it cannot determine every individual's circumstances and that some people operating scams may themselves be exploited.
User signals and limits
Guaranteed returns, secrecy, expiring bonuses, deposits required before withdrawal and fines demanded over messaging apps are reasons to stop and verify through an institution's independently obtained official contact details. AI-generated IDs and transfer screens should not be trusted alone.
The network's scale and losses have not been independently established. OpenAI says conversations suggested hundreds of targets and individual losses of thousands of dollars but explicitly could not verify those claims.