A.I. Undercover: A Closer Look at a Chinese Surveillance Tool
MORE to file under "Chinese Espionage".
Recent findings from OpenAI reveal that a Chinese security operation has developed an AI-powered tool designed to monitor anti-Chinese posts on Western social media. The discovery came to light when a developer, using OpenAI’s technology to debug code, inadvertently provided researchers with key insights into the tool—now dubbed “Peer Review.”
Ben Nimmo, a principal investigator at OpenAI, explained that this is the first instance of an AI-powered surveillance tool of its kind being uncovered. While AI has the potential to combat issues like cybercrime and disinformation, it also carries risks when used for surveillance and other malicious activities.
Interestingly, the tool appears to be built on Meta’s open-sourced Llama technology, which allows developers worldwide to build upon its framework. OpenAI’s report further highlights another campaign called “Sponsored Discontent,” where their technology was used to generate posts critical of Chinese dissidents, translate content into Spanish, and distribute it in Latin America. Additionally, a suspected operation from Cambodia was found using AI-generated comments to fuel a scam known as “pig butchering.”
These developments underscore the complex role AI plays in today’s digital landscape—serving both as a tool for innovation and a means for exploitation. As this technology continues to evolve, so too will the conversations around its ethical and regulatory implications.