OpenAI Bans Cambodian Scam Network Using ChatGPT for Romance and Investment Fraud
OpenAI has banned a coordinated network of ChatGPT accounts based in Poipet, Cambodia, which were used to facilitate romance, investment, and gambling scams. The scammers utilized the AI model to generate fake personas, translate messages, and draft internal communications. This case highlights the growing trend of threat actors leveraging Large Language Models (LLMs) to scale and refine social engineering attacks, making scams more convincing. It also underscores the importance of cross-platform threat intelligence sharing, as the initial lead came from WhatsApp. The network generated content linked to human trafficking and forced labor, and lured victims into fraudulent cryptocurrency and gold trading schemes. OpenAI has shared the threat intelligence with industry partners to prevent further abuse across different platforms.
## BACKGROUND
Large Language Models (LLMs) are increasingly targeted by cybercriminals to automate and polish phishing and chat-based social engineering attacks. OpenAI regularly publishes threat intelligence reports detailing how malicious actors, including state-affiliated groups and organized crime networks, attempt to abuse its AI models.