OpenAI banned a cluster of ChatGPT accounts used in "Operation False Witness," a recovery scam impersonating law firms, real attorneys, and the FBI's Internet Crime Complaint Center (IC3). The operation, likely originating in Cambodia and aligning with Chinese-led criminal networks, used AI to generate content for at least six fake law firms promoted via social media and online ads. Scammers created lawyer personas with lifted or AI-generated profile pictures, reused identities across firms, and directed victims to Telegram for private messaging.

They falsely claimed supervision by the International Criminal Court and promised no upfront fees, then demanded a 15% "service fee," deposits, and consultation fees paid in cryptocurrency. ChatGPT was heavily used to translate messages into "American English" or legalese, and to fabricate credibility documents like fake bar association membership cards and confidentiality agreements.

OpenAI investigators found scammer inputs suggesting individual victims lost thousands of dollars, though independent verification was not possible. The FBI and other authorities were notified. This case highlights how generative AI can scale social engineering by producing polished, multilingual impersonations and forged credentials at low cost.