OpenAI has taken action against a network of ChatGPT accounts involved in a scam operation, dubbed “False Witness,” which very likely originated in Cambodia. This operation used OpenAI's models to create fraudulent content and impersonate trusted entities, including law firms and U.S. law enforcement, to target individuals who had previously been victims of fraud. The details of this disruption were first published in OpenAI’s February 2026 report on malicious AI uses.
The scammers leveraged ChatGPT to support a fraudulent scheme centered on fake law firms and the impersonation of real attorneys and the FBI’s Internet Crime Complaint Center (IC3). This activity aligns with public reporting on Chinese-led criminal scam operations in Cambodia.
Key Points
- OpenAI banned ChatGPT accounts that very likely originated in Cambodia.
- The accounts used AI to impersonate recovery services, law firms, and authorities.
- The scam targeted individuals who had previously been victims of fraud.
- ChatGPT was used to generate content for at least six fake law firms and social media promotions.
- Scammers used OpenAI models to create comments emulating the tone of specialized attorneys.
- The operation involved impersonating the FBI’s IC3 unit and real attorneys.
- ChatGPT was used to draft messages requesting fees, deposits, and cryptocurrency payments from targets.
- Fake attorney registration records, bar association membership cards, and confidentiality agreements were created using OpenAI models.
Context
According to OpenAI Safety, the scam operation used ChatGPT to generate promotional content for social media, craft cold outreach messages, and translate communications with targets on private messaging apps. The scammers frequently requested replies in “American English” or in the style of a lawyer. OpenAI’s policies explicitly prohibit using their tools for fraud or scams.
Why It Matters
This case highlights how AI models can be misused for sophisticated impersonation and fraud, underscoring the need for vigilance among users and robust detection and disruption efforts by AI developers. It also demonstrates the potential for AI to enhance the credibility of fraudulent schemes, making them harder for victims to identify.
What To Do
- Note that OpenAI’s policies strictly prohibit using their tools for fraud or scams.
- Watch for public alerts from law enforcement, such as the FBI, regarding similar scam operations.
- Be aware that AI models can be used to generate convincing but fraudulent documents and communications.
- Exercise caution when encountering unsolicited offers for scam recovery services, especially those requesting advance fees or cryptocurrency payments.
Keep Exploring
/atlas/claude-family /atlas/gpt-family /techniques/system-user-separation