By Shelesh Joshi and Rajiv Menon 
New Delhi, June 11: In a documented case of China propaganda through ChatGPT, China-linked operators covertly used OpenAI’s own platform to manufacture opposition to American trade and technology policy deploying the United States’ most advanced AI tool against the country that built it. The disclosure came in OpenAI’s official June 2026 Threat Report, published on June 10, 2026, on OpenAI’s website, openai.com.
Quick Summary
- OpenAI banned 2 clusters of ChatGPT accounts “likely originating from China” after they ran covert influence operations targeting American domestic debates, according to the OpenAI June 2026 Threat Report.
- The “Data Center Bandwagon” campaign falsely claimed AI data centres were raising electricity bills for ordinary American families and was traced to a private Chinese tech firm holding provincial government contracts, per the official OpenAI report.
- The “Tech and Tariffs” campaign generated political cartoons targeting only President Donald Trump operators instructed ChatGPT to exclude China’s leader Xi Jinping from all content, according to the OpenAI June 2026 Threat Report.
- A linked network on X (Twitter) spread entirely false claims that ChatGPT user data had been compromised, confirmed directly by OpenAI in its official report.
- Both operations rated Category One on OpenAI’s Breakout Scale the lowest possible impact level with no authentic engagement and no evidence of spread beyond their own activity, per the official OpenAI report.
- One operation also requested that ChatGPT design an AI surveillance system to scrape social media for “harmful” content from “key persons” a request OpenAI’s models declined to fulfil in full, according to the official report.
What OpenAI’s Official Report Says About China Propaganda Through ChatGPT
According to the OpenAI June 2026 Threat Report, the company identified and banned two distinct clusters of ChatGPT accounts described as “likely originating from China” accounts that were actively running China propaganda through ChatGPT to manipulate American domestic debates. The accounts used the platform to support what OpenAI called “apparent covert influence operations” targeting domestic American debates on AI infrastructure and tariff policy.
OpenAI named the two campaigns internally the first as the “Data Center Bandwagon” campaign, and the second as the “Tech and Tariffs” campaign. Both clusters used VPNs to access ChatGPT from China, since OpenAI does not permit access to its models from within China. Both prompted ChatGPT in Simplified Chinese while requesting English- and Chinese-language outputs.
Ben Nimmo, Principal Investigator at OpenAI, told journalists, as reported by Reuters: “The operations seemed geared toward manipulating a legitimate debate about American AI and wider American tech policies. Under the circumstances it’s particularly ironic that they tried to use American AI to do it.”
The “Data Center Bandwagon” Campaign
According to the OpenAI June 2026 Threat Report, this first cluster was operated by individuals likely working at a private Chinese technology company conducting work for Chinese provincial-level government clients. OpenAI described this as consistent with “a commercial ecosystem that supports Party-state priorities in public opinion guidance.” The operation’s primary target was US residents. The accounts generated English-language social media comments and images claiming that AI data centre construction was increasing electricity demand and driving up costs for ordinary American families.
Per the OpenAI report, operators asked ChatGPT to produce comic strips based on a real regional newspaper’s reporting about power grid capacity auction prices, framing rising electricity costs as a direct consequence of data centre power demand.
The AI-generated content was posted on X by what the OpenAI report describes as “a set of likely inauthentic accounts,” alongside links to legitimate news stories about power grids and data centre energy demand. Hashtags used, according to the report, included #capacityauction, #datacentersuccess, and #datacenters. The same operation also targeted a second audience: overseas Chinese communities.
According to the OpenAI report, operators asked ChatGPT to generate content insulting Chinese dissident Li Ying, also known as “Teacher Li,” targeting his team’s X account @whyyoutouzhele. OpenAI noted its models refused to generate inflammatory or personal attacks against Li. Other Chinese political commentators targeted for harassment, per the report, included Lu Yiheng, Xu Chi, and the X account @SydneyDaddy1.
The OpenAI report identifies another notable tactic: posing online as US-based Chinese immigrants described as workers, students, mothers, clerks, and investors to encourage a US-based former Chinese police officer to criticise America publicly.
OpenAI described this as “a novel tactic of using fabricated US-based and Chinese immigrant personas to encourage an influencer’s content criticising the US.” Beyond content generation, the OpenAI report states the accounts used ChatGPT to automate their social media workflow including generating code to automate account logins and manage interactions across multiple platforms simultaneously.
The “Tech and Tariffs” Campaign
The second operation, described in the OpenAI June 2026 Threat Report, generated short comments and political cartoons criticising US tariffs and framing Washington as attempting to dominate global technological competition. According to the report, operators framed this competition around tariffs, rare earths, AI, 5G, new energy, and industrial resilience.
One of the most revealing details: operators specified in their ChatGPT prompts that all cartoons should only depict President Donald Trump and “should not include any imagery of China or China’s leader Xi Jinping.” Reuters reported that the cartoons depicted Trump behaving disruptively on the global stage swinging a hammer at a wall labelled “Global Future” or sawing apart a ladder he was standing on.
The OpenAI report states this cluster also generated large batches of short comments in Chinese, designed to “attack the US and Israel” and “amplify anti-Jewish tropes.” Operators additionally asked ChatGPT to propose an AI surveillance system designed to automatically scrape “harmful” information from social media, store logs, download videos for semantic analysis, and send “risk notifications.”
According to the OpenAI report, its models generated only a limited general response and “did not provide ideas for how to collect that data for surveillance purposes.” The OpenAI report flags the language of the prompts as particularly significant. Per the report, prompts “repeatedly used terminology consistent with individuals associated with China’s public security system,” including requests for public opinion risk assessments on protests, crowd movements in Shanghai, and petitioning activity.
One user, the report states, described the accounts it operated using the Chinese term “水军” (water army) a term for coordinated accounts that flood platforms with criticism and instructed ChatGPT to generate content that would “benefit the PRC or advance pro-PRC narratives.” OpenAI states it “could not establish the operators’ precise institutional affiliation,” but concluded the accounts were “likely supporting activities aligned with the interests of the CCP.”
A Third Layer: False Claims About ChatGPT
The OpenAI June 2026 Threat Report identifies a third layer of activity beyond the two named campaigns. A network of likely inauthentic X accounts assessed as likely part of the same broader operation posted false claims that ChatGPT user data had been compromised and that users’ “lives had been negatively impacted.” The OpenAI report states unambiguously: “These allegations were entirely false.”
According to the report, these accounts were created in late 2025, had few or no followers, and had posted only a small number of times. OpenAI found the accounts repeatedly interacted with and amplified content from the “Tech and Tariffs” cluster, and that accounts from both clusters quote-tweeted the same unrelated verified account within hours of each other.
The report also notes content overlaps with a previously identified PRC-origin operation called “Nine-emdash Line”, discussed in OpenAI’s October 2025 threat report. OpenAI says this is “insufficient to conclusively establish a connection,” but that it “reinforces the impression of a nexus of activity on X amplifying influence efforts from China.”
How Effective Were the Operations?
Using the “Breakout Scale” OpenAI’s internal metric for real-world impact the company assessed both campaigns as Category One: activity spanning one platform, with no evidence of breakout, per the official report. The OpenAI report states most social media posts generated “little or no observable engagement.”
It adds: “We found no evidence that the false claims about ChatGPT user data being compromised were amplified by authentic high-reach accounts or beyond the X platform.” However, OpenAI was deliberate in explaining why limited reach does not mean limited significance.
The report states: “The targeting of OpenAI and US data center buildouts is significant not because the operation appears to have shifted public opinion, but because it shows PRC-origin influence operators testing narratives against AI infrastructure a foundation of US technological leadership, economic growth and the broader democratic AI ecosystem.”
The OpenAI report further explains that foreign influence operations “have long sought to latch onto existing local issues and sincerely held beliefs, using them to build credibility, amplify divisions or exacerbate public distrust.” In this case, according to the report, the operators “attempted to covertly insert themselves into an ongoing American debate about the future of the country’s AI capabilities while hiding who they were and what motivated them.”
That distinction matters: the campaigns did not manufacture a debate. They hijacked one that already existed. The report draws a parallel to the 2022 Spamouflage/DRAGONBRIDGE campaigns against rare earths companies, which the Australian Strategic Policy Institute (ASPI) and cybersecurity firm Mandiant assessed as attempts to harm competitors to China’s dominance in that sector.
OpenAI states: “We appear to have observed similar tactics to harm OpenAI’s reputation, albeit unsuccessfully and in an industry where the US is leading.” The timing cited in the OpenAI report is also notable: the campaign against OpenAI followed the CCP’s 15th Five-Year Plan elevating AI as a strategic national priority placing these operations within a deliberate framework of protecting Chinese competitive interests in the AI sector, not merely reacting to US policy.
Why These Targets? The Five-Year Plan Context
The OpenAI June 2026 Threat Report offers precise strategic context for the timing of the campaigns. Per the report, the operations coincided with “a sharp escalation in US-China economic and technology competition when President Trump announced an additional 100 percent tariff on Chinese goods.”
The report also notes the campaigns “followed the Fourth Plenary Session, at which the CCP adopted recommendations for the 15th Five-Year Plan that elevated AI as a strategic technology and industrial priority, calling for accelerated AI innovation and a nationwide ‘AI+’ initiative.” OpenAI draws a parallel to the 14th Five-Year Plan (2021), which “securitised the strategic mineral resources industry and specifically identified ‘high end rare earth functional materials’ as a key priority.”
In both cases, the report concludes, inauthentic accounts targeted private companies in democratic countries operating in sectors Beijing viewed as important to national security and development.
US Congress Responds
Reuters reported that Representative John Moolenaar, Chairman of the House Select Committee on the Chinese Communist Party, responded directly to the OpenAI disclosure on June 10. Moolenaar was cited by Reuters as saying: “Unfortunately, the Chinese Communist Party exploits our openness and works to divide Americans through its United Front organisations and other entities.”
Reuters also reported that Moolenaar acknowledged genuine public concerns exist around data centres noting that more than a dozen US states have introduced or are weighing restrictions on their construction but argued Beijing was exploiting those anxieties rather than engaging with them honestly.
Beijing’s Official Response
Reuters reported that the Chinese Embassy in Washington said it was not familiar with OpenAI’s research. The Embassy was cited by Reuters as saying it “firmly opposed any groundless attacks or smears against China” and that Beijing was working to “ensure AI is a force for good and for all.” The response follows a pattern consistent with Beijing’s public posture on prior influence operation disclosures.
China has repeatedly denied involvement in state-sponsored digital influence campaigns even as technical evidence from multiple technology companies and research institutions has continued to accumulate. The denial also reflects a broader Chinese government position: that accusations of AI misuse by Western tech companies are themselves a form of political pressure rather than objective security reporting. The Chinese Embassy’s statement is the only official Chinese government response in the public record at the time of publication.
A Broader Pattern of Misuse
The OpenAI June 2026 Threat Report places both campaigns within a longer documented pattern of Chinese misuse of its platform. Per the report, a prior OpenAI threat report disclosed the disruption of an individual associated with Chinese law enforcement who used ChatGPT to plan “cyber special operations” targeting Japan’s prime minister, harass dissidents, impersonate Americans, and deploy inauthentic accounts across social media.
The OpenAI report also states that Chinese government entities asked ChatGPT to help draft a proposal for an “early warning” system to track the travel of people categorised as “Uyghur-related and high-risk.” The report notes these disclosures are part of a consistent strategic posture using AI tools to surveil, suppress, and shape narratives both domestically within China and across foreign information environments.
The June 2026 Threat Report is also not OpenAI’s first disclosure of this kind. The company has now published multiple threat intelligence reports documenting how state-affiliated actors from China, Russia, Iran, and others have attempted to exploit its models. What the June 2026 report marks, however, is a notable escalation in strategic specificity: the operations are no longer targeting broad geopolitical themes but are drilling into highly localised American domestic debates energy pricing, infrastructure policy, trade economics.
That shift, from global messaging to targeted local anxieties, is a maturation of the technique. On the choice of using American AI rather than Chinese models, OpenAI states in the official report: “We are not in a position to determine what drove this choice; as we reported in February, China’s strategy of ‘cyber special operations’ emphasises the use of locally deployed Chinese open-weights models.”
The report concludes that both campaigns “attempted to connect US technology policies and industries to everyday economic anxieties and geopolitical instability” themes OpenAI says “are likely to remain attractive for influence operations originating from China.”
The Bigger Picture: What This Case Actually Tells Us
The immediate facts of this story are almost secondary to what the structure of the operation reveals. Beijing did not invent opposition to data centres or tariffs. Those debates are real. The anxieties behind them are genuine. Millions of Americans hold them sincerely.
What the operators did was more calculated: they identified real fault lines in a democratic society and quietly applied pressure using the language of ordinary citizens, on platforms those citizens trust, with content designed to be indistinguishable from organic opinion.
That is the template. Not fabricated crises. Not wild conspiracy. Just real grievances, amplified by foreign hands pretending to be local voices. And this is what makes it harder to counter than any previous generation of propaganda: it does not need to be believed by millions to work.
It only needs to make a small number of authentic voices louder, a legitimate debate slightly more toxic, or a policymaker’s inbox slightly more flooded and the operation has achieved something. The OpenAI report confirms the campaigns failed this time. Category One impact. No breakout. No authentic engagement.
But failure in 2026 is also a rehearsal for 2027.
The operators now know which narratives pass platform moderation. They know which prompts ChatGPT will refuse. They have tested which hashtags land, built a Facebook evasion playbook, and developed a cross-platform amplification architecture all refined through this operation.
The most important line in the entire OpenAI report is not about what these campaigns did. It is about what they were designed for. OpenAI states: “Both clusters attempted to connect US technology policies and industries to everyday economic anxieties and geopolitical instability. These themes are likely to remain attractive for influence operations originating from China because they can be inserted into legitimate public debates while nudging audiences toward distrust of US institutions, technology companies and democratic policy choices.” Read that again slowly. The weapon is legitimacy itself. The target is not a fact it is trust.
For India, this is not a distant American problem.
India is building one of the world’s largest AI infrastructure footprints. It hosts one of the world’s largest internet user populations. It has its own active debates about data centres, energy demand, and technology sovereignty debates that are real, legitimate, and exactly the kind this operational template is designed to exploit.
India is also positioned at a sensitive midpoint in the US-China technology rivalry, which means it is a credible target for influence operations from multiple directions not only as a venue for domestic manipulation, but as a country whose public opinion on China, the US, and technology policy is itself strategically valuable to both sides.
The OpenAI case offers India something rare: a documented, technically detailed, officially sourced manual of exactly how these operations are built and run before they arrive. The question for Indian institutions, platforms, and policymakers is not whether this template will be adapted for use here. Based on the documented pattern, it almost certainly will be. The question is whether India is reading the manual in advance.
Stay ahead with Hindustan Herald — bringing you trusted news, sharp analysis, and stories that matter across Politics, Business, Technology, Sports, Entertainment, Lifestyle, and more.
Connect with us on Facebook, Instagram, X (Twitter), LinkedIn, YouTube, and join our Telegram community @hindustanherald for real-time updates.
Tracking world politics, global markets, trade movements, policy decisions, and the changing balance of economic power.
- Shelesh Joshi
- Shelesh Joshi
- Shelesh Joshi
- Shelesh Joshi
Specializes in South Asian geopolitics and global diplomacy, bringing in-depth analysis on international relations.
- Rajiv Menon
- Rajiv Menon
- Rajiv Menon
- Rajiv Menon