OpenAI has banned several accounts linked to Chinese users who misused ChatGPT to develop code for a social media surveillance tool. This marks the first instance of the company identifying such an operation.
The initiative, dubbed Peer Review, involved users prompting ChatGPT to create sales pitches for a program aimed at monitoring anti-Chinese sentiment across platforms like X, Facebook, YouTube, and Instagram. The operation focused on identifying calls for protests against human rights violations in China, intending to relay this information to Chinese authorities.
OpenAI noted that the accounts exhibited behavior typical of mainland Chinese business hours, utilized Chinese prompts, and engaged with the platform in a manner suggesting manual input rather than automation. The operators reportedly used ChatGPT to verify claims about their findings being communicated to Chinese embassies and intelligence agents monitoring protests in various countries, including the United States, Germany, and the United Kingdom.
Much of the code for the surveillance tool appears to be derived from an open-source version of Meta’s Llama models. Furthermore, the group allegedly employed ChatGPT to draft performance reviews and phishing emails for clients in China.
In a separate incident, OpenAI also banned an account that generated social media posts critical of Cai Xia, a Chinese political scientist in exile in the US. This group also created articles in Spanish that criticized the US, which were published by mainstream news outlets in Latin America, often attributed to individuals or Chinese companies.
For more details, visit the original article here.