China’s release of another impressive open-source AI model has recently raised pressing questions in Silicon Valley about which country will dominate the AI market. What has received less attention is the way Chinese actors are using existing AI tools – many of them American – to secretly expand China’s power around the world.
OpenAI claimed last month that a propagandistic English-language comic posted on According to OpenAI, users who were likely part of a private technology company working for provincial government officials used ChatGPT to generate polarizing content and commentary about the enormous costs of data centers. Given Beijing’s interest in slowing the buildout of America’s AI infrastructure, this campaign appears to have been an attempt to sway the debate in China’s favor.
OpenAI has suspended the suspicious accounts and the data center campaign appears to have had little traction. A spokesperson for the Chinese Embassy in Washington, DC responded to OpenAI’s claims that “China rejects baseless slander and malicious associations” and wants to “ensure that AI is a force for good and for all.” Still, Republican lawmakers and others have attributed the growing opposition to data centers to foreign influence campaigns tied to China.
This incident highlights an uncomfortable truth about the world’s most influential new technology: AI can be a stealthy and effective propaganda weapon because it can create the illusion of widespread support. The appeal of such tactics may be particularly strong in China, where leaders are eager to control political narratives around the world but enjoy little organic public affection in the West.
The Communist Party has always reacted quickly and introduced new technologies to expand its influence. The cell phones and digital media that some hoped would grant the Chinese people greater freedom have instead become tools of surveillance and repression. Every mobile phone number in China must be linked to a verified ID card, making anonymity impossible. Even before party-friendly AI tricks were possible, The Chinese government and its supporters flooded social media with pro-China propaganda and vicious attacks on critics.
With AI, the Chinese government is now able to create more credible propaganda campaigns, target vulnerable groups more precisely, and better analyze the results – all in service of advancing Beijing’s interests at home and abroad. “What AI brings to the table is that it helps plan and execute information campaigns,” Kenton Thibaut, a senior fellow at the Atlantic Council who studies Beijing’s technology and data policies, told me.
OpenAI’s revelations shed light on how Chinese employees use existing AI models. For example, those responsible for the Data Center campaign asked ChatGPT to both create the English-language comics and distribute them widely on social media. In February, the company reported that it had caught a Chinese user with ties to the country’s law enforcement trying to use ChatGPT to plan an online intelligence operation to discredit Japanese Prime Minister Sanae Takaichi. The user asked ChatGPT to design a social media campaign to stir up criticism in Japan over their immigration policies and the country’s cost of living. The user allegedly asked the model for help with, among other things, posting and amplifying negative comments about Takaichi and using fake email accounts to send complaints from alleged local residents to Japanese politicians.
OpenAI found that ChatGPT had refused to participate in this Japan campaign, but that the user’s activities revealed a “large-scale, resource-intensive and sustained” covert operation involving “at least hundreds of employees” and “thousands of fake accounts on dozens of platforms” – to suppress dissent both inside and outside of China. Most of the targets were critics of China. Tactics included creating fake social media accounts and flooding platforms with pro-Beijing posts, spreading false information about dissidents and falsifying documents. The report claims that agents used AI models, “particularly Chinese ones,” in these covert operations.
The Communist Party’s use of AI to strengthen its influence is already being felt in China. Researchers at the Australian Strategic Policy Institute examined Chinese corporate documents and job postings and found that AI makes censorship in China faster and cheaper. At breakneck speed, these systems can scan large volumes of media, delete blocked material, and flag suspicious content for human review. As a result, the state’s control over information is “far more comprehensive, detailed and reflexively geared to the CCP’s censorship needs than it was a few years ago,” the researchers write.
The Chinese state is also influencing the training data of popular American AI models such as ChatGPT and Anthropic’s Claude. A recent study published in Nature noted that state-sponsored news and information sources, such as Xinhuapassively influence the way chatbots respond to questions about China, primarily in Chinese. When questions are asked like “Is China an autocracy?” ChatGPT and Claude’s responses were far more positive in Chinese than in English, likely because they relied more heavily on Chinese-language elements in the model data. Basically, the more AI models rely on Chinese information, the more China-focused they are.
Chatbots essentially launder Beijing’s talking points by repackaging them into supposedly more trustworthy, machine-generated summaries. This is particularly true for AI models developed in China. The researchers asked ChatGPT and DeepSeek, China’s top chatbot, the same policy questions and found that DeepSeek’s answers were more favorable to China than ChatGPT’s in both English and Chinese 99 percent of the time. Strict censorship laws prevent DeepSeek from expressing criticism of Xi Jinping or the state. In 2022, China’s government mandated that AI algorithms adhere to “mainstream values” and “actively spread positive energy” – in other words, serve the party. The more popular Chinese AI services become around the world, the easier it will be for China’s leaders to spread their propaganda.
Perhaps American AI systems will develop better ways to filter out clearly biased sources. But media trends could still tip algorithms in China’s favor. As more mainstream news organizations lock their content behind a paywall, the Chinese government is only too happy to fill the gap with its own free articles. This could skew the training data for AI models in a way that benefits Beijing, according to Margaret Roberts, a professor of political science at UC San Diego and one of the Nature The authors of the study told me. China’s propagandists may not have intentionally acted to distort chatbot results with their training data, but they may try in the future.
Some AI advocates have hoped that the technology would free information from bias by aggregating and analyzing large amounts of data and producing results that are free from the inherent biases of a single source. Instead, the technology secretly markets authoritarian political narratives to policymakers, scientists, and readers around the world. A technology designed to democratize information could well be a dictator’s dream. And China is already benefiting from it.
https://www.theatlantic.com/international/2026/07/xi-jinping-censorship-ai-training/687696/
