OpenAI disrupts scam network targeting hundreds

OpenAI says a Cambodia-based scam operation used ChatGPT across 4 scam types and may have interacted with hundreds of targets. The company’s threat report says conversations referenced individual losses of thousands of dollars; OpenAI banned the associated accounts and shared indicators with partners and authorities, while noting signs that some operators may themselves have faced trafficking or forced labor. The operator lesson extends the case for continuous AI red-teaming: abuse monitoring must correlate identities and behavior across workflows, not assume one account will stay inside one fraud script.