OpenAI said in an official post on its website yesterday that the company had successfully disrupted a scam network based in Cambodia. The scammers used ChatGPT to carry out investment, romance, and gambling scams.
The case occurred earlier this year and was initially brought to light by WhatsApp. The scammers operated as an organized group, flexibly using various narratives, fake identities, and attack methods to deceive victims as effectively as possible.
According to OpenAI's observations, these users also generated content involving human trafficking and forced criminal activities, consistent with numerous public reports. As a result, OpenAI banned a group of coordinated ChatGPT accounts. These users were highly likely to be from Cambodia and operating in the Poipet area of Banteay Meanchey Province.
At the same time, the scam network used ChatGPT models to create fake identities, translate scam messages, produce promotional content for scam operations, and use GPT to draft internal notices and translate employee communications.
The scammers also used fake identities to build relationships of trust with victims, then lured them into fraudulent investment schemes involving cryptocurrency or gold trading. Although the scam scripts varied, users in the network consistently exhibited the same patterns of fraudulent behavior.
OpenAI has now banned the ChatGPT accounts of the users mentioned above and shared relevant information with industry partners. It is still unclear how much financial loss the network caused. Some scammers mentioned in chat windows that some victims had consequently lost thousands of dollars.