OpenAI Discovers AI-Based Chinese Surveillance System
OpenAI, a prominent artificial intelligence firm, announced on Friday that it has discovered an AI-enabled surveillance program created by a Chinese security operation. The tool is reportedly designed to monitor and report anti-Chinese sentiments shared on Western social media platforms.
OpenAI researchers came across this operation, which they named 'Peer Review', when someone involved in the project used OpenAI's technology to debug parts of the underlying computer code. This marked the first instance OpenAI has detected an AI-enabled surveillance mechanism of this type, according to principal investigator Ben Nimmo.
Nimmo observed that malicious actors sometimes unintentionally reveal their activities on the internet via the way they utilize AI models. While the potential misuse of AI for surveillance, hacking, and disinformation is a growing concern, Nimmo and his colleagues also emphasize that AI can be instrumental in identifying and mitigating such harmful behavior.
The investigators believe that the Chinese surveillance tool is based on 'Llama', an AI technology developed and open-sourced by Meta, making it accessible to software developers around the world.
In a comprehensive report on AI's misuse for deceitful and harmful purposes, OpenAI disclosed another Chinese initiative called 'Sponsored Discontent'. This campaign reportedly used OpenAI's technology to create English posts criticizing Chinese dissidents. The same group has also used OpenAI's resources to translate articles into Spanish and distribute them in Latin America, with content critical of US society and politics.
In a separate discovery, OpenAI researchers detected a campaign, presumably originating from Cambodia, that used their technology to create and translate social media comments promoting a fraudulent scheme known as "pig butchering". The AI-generated comments were allegedly used to entice men online into an investment scam.
In a separate matter, OpenAI and Microsoft have been accused of copyright infringement related to AI systems by The New York Times. Both companies have denied these allegations.