OpenAI has banned two groups of ChatGPT accounts linked to China that were used to run covert influence campaigns targeting American debates over technology policy and data center construction.

The first campaign, called "Data Center Bandwagon," used ChatGPT to create social media posts and comic strips arguing that AI data centers are driving up electricity costs for ordinary American households. The operators wrote their prompts in Simplified Chinese and used VPNs to access the system, then posted the content on X while posing as Americans from various backgrounds. The comic strips focused on capacity auction prices set by grid operators, drawing on real reporting from regional newspapers. The posts appeared alongside links to legitimate news stories and used hashtags like #capacityauction.

The second campaign, called "Tech and Tariffs," produced political cartoons and comments critical of US technology policies and tariffs. The operators specifically requested that the cartoons depict President Trump but not Chinese leader Xi Jinping. This cluster also generated content in multiple languages, including English, Italian, Japanese, and Traditional Chinese, with the Traditional Chinese content directed at audiences in Taiwan. One operator referred to the network of accounts as a "water army," a Chinese term for coordinated troll networks, and asked ChatGPT to help design a system for monitoring social media posts from individuals flagged as risks. OpenAI said its model provided only generic data storage advice and declined to assist with data collection. The same network of fake accounts also spread false claims that ChatGPT user data had been stolen, which OpenAI believes was an attempt to damage its reputation.

OpenAI assessed that the Data Center Bandwagon operators were likely a social media team at a private Chinese tech company providing services for provincial-level government clients. The company rated both campaigns at Category One on its Breakout Scale, meaning the activity remained confined to a single platform with no evidence of reaching authentic audiences. Ben Nimmo, OpenAI's principal investigator, stated that the debate over data center energy costs already existed in the United States and that the operation represented an attempt by Chinese actors to interfere in an ongoing American discussion rather than to create a new one.

The real-world context behind the campaign involves PJM Interconnection, the largest grid operator in the United States, which has documented sharp increases in power costs in regions with heavy data center concentration. PJM's independent market monitor has attributed an irreversible 75.5% rise in power costs across its region to data center demand, with wholesale prices near some data center clusters climbing as much as 267% over a five-year period. Three US senators have formally requested explanations from Amazon, Google, and Meta about costs being passed along to residential customers. Research has found that wholesale electricity prices could rise between 6 and 29 percent nationally by 2030 due to data center expansion. A Gallup poll found that 71 percent of Americans now oppose data centers being built near their homes, up from 42 percent just nine months ago. Opposition to data center construction has stalled or halted more than $156 billion in planned projects across 48 sites, according to the research firm Data Center Watch.

Senate Intelligence Committee Chairman Tom Cotton has called on the Department of Justice to investigate what he describes as a covert Chinese influence campaign aimed at undermining AI data center development in the United States. The Arkansas Republican sent a letter to acting Attorney General Todd Blanche requesting a federal probe, stating that maintaining the nation's AI advantage is vital to economic strength, diplomacy, national security, and military power. Cotton's concerns are partly based on a report from the Bitcoin Policy Institute, which found that Chinese state media outlets including CGTN, China Daily, and Global Times, along with Russia's RT, have launched propaganda campaigns targeting US AI data centers and export controls. The report also identified a network linked to Shanghai-based expatriate Neville Roy Singham that has spent five years producing content opposing American AI infrastructure, labs, and export controls. Cotton said congressional investigators have found that Singham has close ties to the Chinese Communist Party and that foreign-funded charitable groups have funneled more than $2 billion into American advocacy groups, with a significant portion going to organizations promoting opposition to data centers. Cotton acknowledged that Americans have valid concerns about data center growth and said he supports common-sense regulation but warned that foreign adversaries must not be allowed to exploit those fears.

The Chinese Embassy in Washington said it had not read OpenAI's report but firmly opposed what it called groundless attacks or smears against China, stating that China wants to ensure AI serves as a force for good.

Separately, Chinese government-linked operatives have been caught rebuilding a botnet of compromised routers and internet-connected devices. The botnet, known as the JDY cluster and linked to the threat group Volt Typhoon, has grown to more than 1,500 compromised devices after the FBI dismantled a related cluster in January 2024. Lumen's Black Lotus Labs reported that the operators focus on exploiting newly disclosed vulnerabilities quickly, with the US military and related entities being the most frequently targeted. The security team urged all organizations to follow guidance from CISA and the UK's NCSC to defend against these covert networks.

The US Justice Department also announced it had seized 13 fake consulting company websites used to target current and former US security clearance holders. Since November 2023, these sites and associated job postings on platforms like LinkedIn advertised consulting positions as a cover to recruit and bribe individuals for classified and sensitive government information. The alleged conspirators pressured candidates to share confidential reports and paid them using cryptocurrency and online accounts under fictitious names to conceal their identities.

OpenAI compared the campaigns to the 2022 Spamouflage operation, which targeted rare earth companies after China's 14th Five-Year Plan prioritized that sector. The new activity followed the adoption of the 15th Five-Year Plan, which elevated AI as a strategic industry for China. The campaigns highlight a broader pattern of foreign actors using American AI tools to attempt to shape domestic policy debates, even when the underlying concerns they amplify are based on real issues affecting American communities.

Original Sources: 1, 2, 3, 4, 5, 6, 7, 8 (openai) (chatgpt) (china) (gallup) (investigations)

This article provides limited practical value to a normal person. It reports on a specific claim by OpenAI about a foreign influence campaign but does not offer actionable steps, tools, or choices that a reader can use. There are no resources mentioned, no instructions to follow, and no clear actions a person can take based on this information. The article simply recounts what OpenAI says happened without giving the reader anything to do.

The educational depth is shallow. The article states facts about the campaign but does not explain how influence operations work, what laws apply to foreign lobbying in the United States, or why certain social media tactics are effective. The numbers and percentages mentioned, such as the 267 percent potential price increase or the 71 percent opposition figure, are not explained in a way that helps the reader understand how they were calculated or what they mean for everyday life. The information stays on the surface and does not teach the reader anything they can apply to other situations.

Personal relevance is narrow. The story involves a specific company, a specific foreign campaign, and a specific political debate that does not directly affect most readers' safety, money, health, or decisions. While electricity costs and data center construction are general concerns, this article does not connect the event to the reader's life in a meaningful way. It does not explain how likely someone is to be affected by foreign influence campaigns or what to do if they encounter suspicious content online. The relevance is limited to people who follow tech industry news or U.S.-China relations closely.

The public service function is weak. The article does not offer warnings, safety guidance, or emergency information. It does not tell readers how to identify foreign influence campaigns, what to do if they see suspicious social media posts, or how to evaluate online content critically. The mention of Republican lawmakers calling for investigations does not give the public useful advice. The article appears to exist mainly to report a claim rather than to help people act responsibly or stay informed.

There is no practical advice in the article. No steps or tips are given that an ordinary reader can follow. The guidance is entirely absent, not just vague or difficult. The article does not even suggest general precautions a person might take when encountering online content about controversial topics.

The long term impact is minimal. The information does not help a person plan ahead, improve habits, or make stronger choices. It focuses on a short lived news event with no lasting benefit. A reader cannot use this story to avoid problems in the future because no lessons or patterns are explained.

The emotional and psychological impact is neutral to slightly negative. The article does not create fear or shock, but it also does not offer clarity or calm. It simply tells a story that may leave the reader feeling that foreign influence is something that happens in news reports, without any constructive way to think about it. It does not harm, but it does not help either.

There is no clickbait or ad driven language. The article is straightforward and does not use exaggerated or dramatic claims. It does not sensationalize the event or rely on shock to maintain attention. The tone is factual and calm.

The article misses several chances to teach or guide. It presents a problem, foreign influence campaigns using AI tools, but fails to provide context, examples, or ways for the reader to learn more. It could have explained how to identify coordinated inauthentic behavior online, what to do if someone suspects a social media account is not genuine, or how to evaluate claims made by companies with vested interests. A reader who wants to learn more could compare independent news accounts of similar influence operations, look for patterns in social media behavior, or consider general media literacy practices like checking multiple sources before accepting a claim.

To add real value, a reader can take several practical steps based on general reasoning and universal critical thinking principles. When encountering online content about controversial topics, especially content that seems designed to provoke strong emotions, pause and check whether the source is credible and whether other independent outlets are reporting the same story. If a company or organization makes a claim that benefits its own interests, look for independent verification before accepting it as fact. When evaluating statistics or numbers in news reports, ask how they were calculated, what time period they cover, and whether they represent a representative sample or an extreme case. If you see social media posts that seem to repeat similar talking points or use similar language, consider the possibility that they may be part of a coordinated effort rather than organic public opinion. When making decisions about where to live or what to support, rely on multiple sources of information rather than a single report, and prioritize local, verifiable data over distant, abstract claims. These steps are realistic, widely applicable, and grounded in common sense. They help a reader assess risk, choose more reliable sources, and prepare for situations that could affect their understanding of current events, even though the original article offered none of this guidance.

OpenAI has published a report claiming that users based in China ran a covert campaign using ChatGPT to turn American public opinion against data centers in the United States.

This opening uses the word "claiming" which can make the report sound less than certain. It also says "covert campaign" which pushes a feeling of sneaky wrongdoing. The phrase "turn American public opinion" makes it sound like Americans cannot think for themselves. This helps OpenAI look like a truth-teller while making China look like a hidden enemy.

The company says a cluster of now-banned accounts, which it calls the "Data Center Bandwagon" campaign, used the chatbot to generate social media comments and images arguing that AI data center construction was driving up electricity costs for ordinary families.

Giving the campaign a catchy name like "Data Center Bandwagon" makes it sound bigger and more organized than it might be. The word "ordinary" makes regular people seem helpless and victims. This helps OpenAI seem like it is protecting everyday Americans from foreign tricks.

The content was then posted on social media by accounts posing as Americans from various backgrounds.

The word "posing" makes the accounts sound like liars pretending to be someone else. It hides the possibility that some might have been real people with real concerns. This makes the whole campaign seem fake and controlled by outsiders.

OpenAI states the accounts were likely operated by a social media team at a private Chinese technology company working for provincial-level government clients, and that a VPN was used since ChatGPT is not accessible from China.

The word "likely" means they are guessing but it is stated as fact. Saying "working for provincial-level government clients" makes it sound like the Chinese government is behind everything. This pushes a story of government control without full proof.

The central tension in the report is that the core claim pushed by the campaign, that data centers are raising electricity prices, is factually accurate.

Calling this a "central tension" is a trick. It makes the reader think there is a big conflict when really the text is saying the campaign told the truth. This hides the fact that OpenAI is admitting the problem is real while trying to focus on who said it.

Multiple independent studies have confirmed that data center expansion is driving significant increases in power costs.

The word "independent" is used to make the studies sound fair and trustworthy. But the text does not name the studies or show who paid for them. This is a trick to make the reader accept the claim without checking.

Research has found that wholesale electricity prices could rise between 6 and 29 percent nationally by 2030, with some areas near data centers seeing price spikes as high as 267 percent.

The word "could" makes this a guess but it is presented as research findings. The huge number of 267 percent is picked to shock the reader. This pushes fear without saying how likely it is to happen.

A separate Gallup poll found that 71 percent of Americans now oppose data centers being built near their homes, up from 42 percent just nine months ago, suggesting the opposition exists independently of any foreign influence effort.

This poll is used to show that real Americans already oppose data centers. But the jump from 42 to 71 percent in nine months is very fast and the text does not explain why. This helps OpenAI say the foreign campaign did not matter much.

OpenAI says the campaign had little measurable impact on public opinion.

This is OpenAI saying their own story had no real effect. But they are the ones who found the campaign so they get to say how big or small it was. This helps OpenAI look honest while also making the threat seem small.

Critics have noted that OpenAI has a vested interest in downplaying anti-data center sentiment, given the company's central role in the AI industry.

The word "vested interest" is a soft way of saying OpenAI might be lying to protect itself. But the text puts this in the mouths of "critics" so OpenAI does not have to say it directly. This is a trick to show both sides while still pushing doubt.

Beyond electricity costs, communities have raised legitimate concerns about data centers' environmental impact, heavy water consumption, use of valuable land, and broader resentment toward AI technology.

The word "legitimate" is used to admit these concerns are real. But this also helps OpenAI look fair and balanced. It hides the fact that OpenAI profits from the same AI growth that causes these problems.

Opposition to data center construction has stalled or halted more than 156 billion dollars in planned projects across 48 sites, according to the research firm Data Center Watch.

This big number is used to show how serious the opposition is. But the text does not say how many of these projects were actually needed or good for people. It just uses the dollar amount to make the backlash seem huge.

Some Republican lawmakers have called for investigations into whether Chinese influence operations are fueling the backlash, though the evidence suggests that rising utility bills and environmental concerns are sufficient explanations for public opposition on their own.

Naming only "Republican lawmakers" makes this a party issue. It hides the fact that others might also want investigations. The phrase "sufficient explanations" is used to push the idea that China has nothing to do with it. This is a political bias that helps one side look reasonable.

The text expresses several emotions that work together to shape how the reader feels about OpenAI's report on the alleged Chinese influence campaign. The strongest emotion is alarm, which appears throughout the piece and serves as the dominant emotional tone. Words like "covert campaign," "posing as Americans," and "foreign influence" create a sense that something sneaky and potentially threatening is happening. The strength of this alarm is moderate to high, particularly when the text describes accounts pretending to be Americans and using a VPN to access ChatGPT from China. The purpose of this alarm is to make the reader feel that this is a serious issue that deserves attention, even if the actual impact turns out to be small. The word "covert" is especially powerful because it suggests secrecy and deception, making the campaign sound like a spy operation rather than just a few people posting online.

Closely related to alarm is a sense of suspicion, which emerges from the way OpenAI describes the alleged operators of the campaign. The text states that the accounts were "likely operated by a social media team at a private Chinese technology company working for provincial-level government clients." The word "likely" is important here because it means OpenAI is not completely certain, but the phrase "working for provincial-level government clients" pushes the reader to believe the Chinese government is behind this. This suspicion is directed at China and serves to make the reader feel that this is not just a random act but something organized and possibly government-backed. The strength of this emotion is moderate, and its purpose is to create a sense of an external threat, which helps OpenAI position itself as a defender of truth and transparency.

A quieter but still meaningful emotion is reassurance, which appears when the text discusses the actual impact of the campaign. OpenAI says the campaign "had little measurable impact on public opinion," which is meant to calm the reader and suggest that the threat is not as big as it might seem. The strength of this reassurance is moderate, and its purpose is to prevent the reader from panicking or overreacting. However, this reassurance is complicated by the fact that OpenAI has a vested interest in downplaying the campaign's impact, since the company benefits from the growth of the AI industry. The text acknowledges this by noting that "critics have noted that OpenAI has a vested interest in downplaying anti-data center sentiment," which introduces a note of doubt about whether the reassurance should be fully trusted.

There is also a sense of validation that emerges when the text discusses the core claim of the campaign, that data centers are raising electricity prices. The text states that this claim is "factually accurate" and that "multiple independent studies have confirmed" it. This validation serves to make the reader feel that the concerns about data centers are real and justified, regardless of who is raising them. The strength of this emotion is moderate to high, and its purpose is to shift the focus from the foreign influence angle to the underlying issue of electricity costs. By confirming that the campaign's main argument is true, the text makes the reader feel that the real problem is not the foreign campaign but the actual impact of data centers on people's lives.

A feeling of concern about the broader implications of data center construction also runs through the text. The mention of "environmental impact, heavy water consumption, use of valuable land, and broader resentment toward AI technology" creates a sense that there are many legitimate reasons to oppose data centers beyond just electricity costs. The strength of this concern is moderate, and its purpose is to make the reader feel that the opposition to data centers is widespread and well-founded, not just the result of foreign manipulation. The statistic that 71 percent of Americans now oppose data centers being built near their homes, up from 42 percent just nine months ago, reinforces this concern by showing that public opinion is shifting rapidly for reasons that have nothing to do with China.

A subtle emotion of frustration appears in the description of how the campaign was carried out. The phrase "posing as Americans from various backgrounds" carries a tone of disapproval, suggesting that the accounts were being dishonest about who they were. This frustration is directed at the alleged operators of the campaign, and it serves to make the reader feel that their actions were wrong and deceptive. The strength of this emotion is low to moderate, but it is effective because it builds on the earlier alarm and suspicion, creating a cumulative sense that the campaign was not just misguided but actively harmful.

There is also a faint note of defensiveness in the way the text handles the political angle. The mention of "some Republican lawmakers" calling for investigations into Chinese influence operations introduces a partisan element, but the text quickly counters this by stating that "the evidence suggests that rising utility bills and environmental concerns are sufficient explanations for public opposition on their own." This defensiveness serves to distance the story from partisan politics and to present the issue as one that transcends political divisions. The strength of this emotion is low, and its purpose is to make the reader feel that the story is about facts, not politics.

The writer uses several tools to increase the emotional impact of the text. One of the most effective is the use of large, dramatic numbers to convey scale. The figures of 156 billion dollars in stalled projects, 267 percent potential price increases, and 71 percent opposition are all designed to shock the reader and make the issue feel enormous. These numbers are not explained in detail or broken down into smaller, more relatable units, which makes them feel almost incomprehensible and therefore more alarming. The purpose of these numbers is to make the reader feel that this is a big deal, even if the foreign influence campaign itself had little impact.

Another tool is the use of strong action words that carry emotional weight. Words like "covert," "posing," "misled," and "stalled" all suggest wrongdoing and harmful consequences. These words are not neutral; they are chosen to make the reader feel that the situation is serious and that someone is at fault. The phrase "posing as Americans" is particularly effective because it implies deception and dishonesty, which increases the emotional impact.

The writer also uses comparison and context to heighten the emotional effect. By describing the campaign as "covert" and then immediately noting that the campaign's core claim is "factually accurate," the text creates a tension that makes the reader feel conflicted. On one hand, the campaign sounds sneaky and wrong. On the other hand, the campaign was telling the truth about a real problem. This comparison serves to complicate the reader's emotions, making them feel both alarmed by the foreign influence and validated in their concerns about data centers.

Repetition of the theme of legitimacy serves as another emotional tool. The text repeatedly emphasizes that the concerns about data centers are "legitimate" and that public opposition exists "independently of any foreign influence effort." By returning to this theme multiple times, the writer creates a pattern in the reader's mind, reinforcing the idea that the real issue is not the foreign campaign but the actual problems caused by data centers. This repetition builds a cumulative emotional effect, making the reader increasingly sympathetic to the anti-data center movement and less focused on the foreign influence angle.

The overall emotional strategy of the text is to alarm the reader about foreign influence while simultaneously validating the concerns that the influence campaign was exploiting. The emotions are arranged in a deliberate sequence, beginning with the shocking revelation of a covert campaign, moving through the confirmation that the campaign's claims were true, and ending with the suggestion that public opposition is driven by real problems rather than foreign manipulation. This structure guides the reader from immediate alarm to a more nuanced understanding, making the story feel both urgent and complicated. The writer achieves this through carefully chosen words, dramatic numbers, and a structure that builds emotional intensity while also introducing doubt and complexity, ensuring that the reader finishes the piece feeling that this is a serious matter but one that requires careful thought rather than simple reactions.