OpenAI Says Fake Accounts From China Tried To Turn Americans Against Data Centers

OpenAI published a report about ChatGPT users, who they say may be based in China, who are using the chatbot to orchestrate a campaign designed to sway Americans’ perceptions of AI data centers. It divided users into two groups, the first of which had chosen the “Data Center Bandwagon” group. The accounts are divided into groups that are said to have asked ChatGPT to produce talking points in English and images, such as comic characters, focusing on how AI data centers drive electricity demand and how that leads to higher consumer bills.

The company says these users impersonated Americans from various social media sites, where they posted text and image output they received on ChatGPT. OpenAI believes they are part of a social media group at a private Chinese company that serves local government clients. Apparently they even uploaded a file to the chatbot explaining their intentions and strategies on how to change public opinion and how to set up fake social media accounts without being detected.

The group has also targeted Chinese people living abroad, going so far as to ask ChatGPT to carry out defamation aimed at harassing Chinese dissidents and political commentators. While posing as US-based Chinese immigrants and professionals, they urged online public figures to talk more about US policy failures, too.

It’s worth noting that while users may have used fake accounts to post on social media, the company says they posted links to “legitimate news stories about the power grid operator’s power auction and data center power demand.” Rising electricity costs in cities and towns near AI data centers is a real concern and one of the first effects of the AI ​​boom that people feel. According to a Bloomberg report, electricity now costs about 267 percent more per month compared to five years ago in areas near data centers, because their power needs to be provided through.

Meanwhile, a second set of accounts that OpenAI found produced comments and images criticizing US taxes and technology policies. They are particularly focused on producing content critical of the US and insist that the country has been stabbing its allies. The group asked ChatGPT to keep Chinese President Xi Jinping in the images they produced and told the chatbot to write comments in English, Italian, Japanese and traditional Chinese to target a Taiwanese audience.

OpenAI acknowledges that the campaigns have failed to achieve much authentic communication and that they have not changed public opinion. Based on OpenAI’s own report, they focus on real issues that have been controversial and widely discussed on the internet since the beginning. The company explains that the reason these campaigns are important is because “employees have tried to insert themselves into America’s ongoing conversation about the future of the country’s AI capabilities while hiding who they are and what motivated them.” As to why the campaigns chose to use an American AI chatbot instead of DeepSeek, even OpenAI could not say. “We are not in a position to determine what led to this choice,” its report reads.