OpenAI bans PRC-linked AI influence ops on US AI debates

OpenAI's report details two banned clusters of ChatGPT accounts that it says likely originated from China. The first, named 'Data Center Bandwagon,' generated comments and images claiming AI data center buildouts increase electricity prices for families. The second, 'Tech and Tariffs,' criticized US tariffs and explicitly instructed content to avoid mentioning China's leader Xi Jinping, instead including only President Trump. This second cluster was also tied to a network of likely inauthentic accounts that spread false claims about ChatGPT user data being compromised.

OpenAI emphasizes that the significance lies not in any apparent shift in public opinion, but in the operators' testing of narratives against AI infrastructure. This infrastructure is central to US technological leadership, economic growth, and the broader democratic AI ecosystem. The campaigns sought to exploit existing public concerns about energy prices and local impacts of data center development, a tactic consistent with how foreign influence operations often latch onto genuine local issues to build credibility.

The report's publication aims to help industry, governments, and civil society identify and disrupt such attempts. The broader stakes involve protecting democratic institutions from covert interference, particularly as AI tools become more accessible. The effectiveness of such operations may hinge on whether they can amplify real grievances, but OpenAI's finding of no meaningful breakout suggests current efforts have had limited reach.