OpenAI removes China-linked users accused of amplifying the US power cost debate

AI story shaking the internet

This story comes from a real OpenAI investigation about online influence campaigns linked to ChatGPT misuse in global debates. It focuses on how AI tools can sometimes be used to generate social media posts and images that look like real public opinion.

The official report described two clusters of accounts that tried to shape conversations about U.S. technology issues, especially energy use and tariffs. OpenAI said these accounts were later removed after internal review.

What does OpenAI uncover?

OpenAI identified two coordinated clusters of accounts believed to be linked to China that used ChatGPT to generate posts and images. The content focused on U.S. debates about AI infrastructure and trade policy.

The company said the accounts used VPNs and fake identities to appear like everyday users on social media platforms such as X.

The official campaign names

The two clusters were labeled Data Center Bandwagon and Tech and Tariffs in OpenAI’s internal investigation. Both were tied to online discussions already happening in the United States.

One group focused on electricity costs linked to data center growth, while the other focused on trade restrictions and global tech competition. OpenAI said both campaigns were removed after detection.

Data center energy claims

The Data Center Bandwagon group created posts suggesting that AI data centers were raising household electricity bills. They used simple language and comic-style visuals to spread the message.

These posts were designed to blend into ongoing public debates about energy use in technology infrastructure. The claims were shared through fake accounts posing as U.S. residents.

Little-known fact: Real-world reports have shown electricity prices rising near some data center regions due to high energy demand, making the topic already widely discussed.

Why did the topic already exist?

The campaign did not create a new issue but used an existing concern about rising energy demand. Many U.S. communities are already debating data center growth.

Experts say this made the content more believable, since it mixed real facts with AI-generated messaging. OpenAI noted that this blending strategy is common in online influence attempts.

The tariffs-focused campaign

The second campaign identified by OpenAI was called Tech and Tariffs. Unlike the data center effort, this operation focused on trade policy, tariffs, technology competition, and international economic issues.

The operators used ChatGPT to generate short comments, political cartoons, and multilingual content designed for audiences across different regions. Investigators found that the campaign produced material in English, Italian, Japanese, and Traditional Chinese.

Some posts criticized U.S. tariff policies and portrayed technology competition through cartoon imagery. OpenAI noted that the operators appeared interested in reaching audiences beyond the United States, especially in regions connected to broader technology and trade discussions.

Little-known fact: OpenAI concluded that while influence attempts exist, this particular campaign had minimal real-world effect on public opinion.

How was ChatGPT used?

According to OpenAI, ChatGPT was used to generate and edit content for online influence activity. The operators created prompts, received AI-generated responses, and then distributed those outputs through likely inauthentic social media accounts.

This allowed them to produce comments, captions, cartoons, images, and talking points on a greater scale. The investigation showed how generative AI can help users quickly create persuasive online material around public debates.

OpenAI said these cases demonstrate how AI systems can be misused by people attempting to influence online discussions while hiding their identity and purpose.

Fake online identities used

A major part of the operation involved creating online personas that appeared authentic. The accounts presented themselves as students, workers, parents, immigrants, and community members. Their posts were designed to look like ordinary opinions rather than coordinated messaging from a central source.

This approach is common in influence campaigns because people are generally more likely to trust content that appears to come from fellow citizens. OpenAI said the operators attempted to hide their true identities through VPN usage and carefully constructed social media profiles.

Low real-world impact

Even though the campaigns generated a considerable amount of content, OpenAI concluded that their overall impact was very limited.

Most posts received little engagement, and investigators found no evidence that the campaigns significantly influenced public opinion or policy discussions. The company rated the operation at the lowest level on its internal Breakout Scale.

This means the content largely remained within the campaign’s own network and failed to attract widespread organic attention. Researchers noted that creating content is only one part of an influence effort; reaching real audiences remains much more difficult.

Why do data centers matter?

Data centers are the backbone of modern digital services. They support cloud computing, streaming platforms, online storage, websites, and, increasingly, artificial intelligence systems. As AI models become more powerful, companies are investing heavily in larger and more advanced facilities.

These projects often bring economic benefits, including construction jobs and technology investment. At the same time, communities sometimes raise questions about electricity consumption, water use, land development, and infrastructure planning. Because of these competing interests, data centers have become a major topic in local and national policy discussions.

OpenAI’s response

After identifying the activity, OpenAI banned the accounts involved and published details through its threat intelligence reporting. The company said transparency is important because it helps researchers, policymakers, and the public understand how AI tools may be misused.

OpenAI also works with other organizations to monitor emerging influence operations and improve safeguards. The company emphasized that exposing these campaigns helps reduce their effectiveness and allows platforms to learn from the tactics being tested. Publishing reports also helps build awareness about evolving online threats.

Bigger challenge online

The case highlights a broader challenge facing the internet today. Generative AI can create large volumes of text and images quickly, making it easier for bad actors to produce content at scale.

OpenAI said similar attempts to scale influence messages may continue in the United States and globally. At the same time, generating content is not the same as changing minds.

Many influence campaigns still struggle to gain authentic engagement, and OpenAI said these campaigns showed little evidence of meaningful breakout. Detection, industry sharing, and platform responses remain important parts of limiting these operations.

Want to see how AI companies are competing to shape the future of these powerful tools? Check out how Anthropic challenges OpenAI in the lucrative professional services market for a closer look at where the industry is heading next.

What happens next?

The investigation offers an early look at how AI may be used in future influence operations. Experts expect that political, economic, and technology-related debates will remain attractive targets because they already involve strong public opinions and active discussion.

For technology companies, the challenge is finding ways to support innovation while limiting misuse. For users, the lesson is to evaluate online content carefully and pay attention to sources, especially when posts seem designed to provoke strong reactions. As AI becomes more common, digital literacy may become even more important than it is today.

Want to see how companies are scaling the infrastructure behind this new wave of AI-driven content? Check out Meta’s exploration of unconventional data center construction for AI growth to explore what’s powering it all behind the scenes.

What do you think about AI being used in online influence campaigns like this? Does it raise concerns for you? Drop your thoughts in the comments, and don’t forget to leave a like.

This slideshow was made with AI assistance and human editing.

Don’t forget to follow us for more exclusive content on MSN.

Read More From This Brand:

• Microsoft faces backlash for 6000 visa push amid job cuts
• Microsoft experiments with Copilot AI that can operate across your desktop
• Microsoft faces scrutiny over alleged Chinese engineers handling DOD systems