OpenAI terminated accounts linked to a Cambodia-based criminal network that used ChatGPT to run investment fraud, romance scams, gambling schemes, and impersonation operations simultaneously.
The details buried in this report matter more than the headline. OpenAI documents exactly how the operators used the model: drafting multilingual scripts, generating fake personas, and producing persuasive conversational content at scale. This is a case study in how LLMs lower the operational cost of organized fraud.
The full post is worth reading for its breakdown of the specific use patterns OpenAI flagged before termination. The harder question it raises but does not fully answer: how many operations it did not catch.
[READ ORIGINAL →]