Foreign influence operators recently turned to an American artificial intelligence tool to convince Americans that the tech is ruining their communities. OpenAI published a comprehensive threat intelligence report revealing it detected and disabled multiple accounts, likely based in China, that used ChatGPT to fuel public anger over local data centers and US tech policies with anti AI-propaganda.
The “Data Center Bandwagon” strategy for anti-US AI propaganda using ChatGPT
According to OpenAI principal investigator Ben Nimmo, the covert operation essentially tried to piggyback onto genuine, pre-existing domestic arguments regarding energy grids and data center power consumption. One specific cluster of accounts, dubbed the “Data Center Bandwagon” group, prompted ChatGPT in simplified Chinese to generate English-language social media comments and political comic strips.
The operators then used fake accounts across networks like X and YouTube, posing as everyday Americans and immigrants. They published content claiming that massive AI data center construction projects are drastically driving up residential electricity bills (via Politico).
It’s true that data center expansion has caused real-world energy rate spikes in certain areas. However, OpenAI noted that the threat actors simply attempted to insert themselves into the conversation covertly while hiding their true identities.
The operators behind the campaign were hardly subtle. In a massive security blunder, they uploaded text files directly into ChatGPT. The files detailed their exact strategies for evading detection, building audience reach, and managing long-term account viability on platforms like Facebook. OpenAI traced this activity back to an unnamed Chinese technology company. This organization actively holds multiple software contracts with regional Chinese governments.
Targeting tariffs and policy failures
A second distinct cluster of banned accounts used ChatGPT to attack US trade restrictions and tech regulations. These prompts explicitly instructed the chatbot to write commentary criticizing the Trump administration’s tariffs. They specifically focused on requesting generated images featuring US President Donald Trump but excluding Chinese President Xi Jinping. The group distributed content across global audiences in English, Italian, Japanese, and traditional Chinese. The images painted the US as a bad partner that backstabs its traditional allies.
Despite the sophistication of using generative models to scale text and imagery, the overall operation was a massive flop. The campaigns did terribly on traditional influence metrics. It failed to gain any authentic audience traction or move the needle on public opinion. Many of the generated materials felt clunky and demonstrated a clear lack of familiarity with American internet culture and viral formatting.
The post China Groups Used ChatGPT to Generate Anti-American AI Propaganda, OpenAI Finds appeared first on Android Headlines.
