OpenAI has disclosed that it recently shut down covert online influence efforts it says were run from China, after the groups allegedly used ChatGPT to generate propaganda about contentious U.S. policy debates.

Early testing of AI for influence

According to OpenAI’s latest investigations, China-linked operators began experimenting with ChatGPT to draft social media content on U.S. issues in late 2025, focusing first on disputes over AI data centers and electricity costs.¹ One campaign, later dubbed “Data Center Bandwagon,” produced comments and comics suggesting that data centers were driving up power bills for American families.²

Investigators say the operators posed as Americans and tried to “jump onto the bandwagon of a genuine pre-existing domestic debate” about data center construction and local impacts, using fake accounts on platforms such as X and Facebook.¹ OpenAI concluded the users were likely part of a social media operations team at a private Chinese tech firm working for provincial-level government clients.¹

Expansion to tariffs and tech policy

A second operation, labeled “Tech and Tariffs,” used ChatGPT to create political cartoons and posts critical of U.S. tech policy and tariffs, including AI-generated images of Donald Trump and attacks on the U.S. “push for global tech dominance.”² These efforts sought to exploit already-heated debates over trade and economic pain from tariffs, rather than create new controversies from scratch.²

Limited impact but growing concern

Across both campaigns, OpenAI says the content was small-scale, short-lived, and gained “little to no authentic engagement,” with “no evidence of meaningful breakout beyond its own activity” and “no signs that they succeeded.”¹ Nonetheless, the company argues the operations show how pro-China actors are “testing AI tools to amplify existing political and economic divisions in the U.S.”²

OpenAI has now banned the implicated accounts and stressed that, while the campaigns flopped, the underlying public debates over data center power costs and tariffs remain very real.³

---

1. Business Insider — "OpenAI says a suspected China-linked influence operation tried to sway the debate about US data centers" f37eb27bb6eb542573eb49dc7d3baa2e09ee022d5029e06440f561d8853923b5

2. Axios — "China-linked operatives used ChatGPT to shape AI data centers and tariff debates" a81a7ca9012c0794b5469398115d2c7a02b7833e21bb77fc19da7d4ae3f9bbd7

3. The Next Web — "OpenAI says China-linked accounts used ChatGPT to fuel US data centre opposition" 91fedb0d3ebcbf7a122b1dd4b5b9c476e88914315289be8bdabbb7997647c0cd