OpenAI Disrupts Chinese Influence Campaigns Targeting US AI Data Centres

I’ll work through each flag before writing the final article.

Issue 1 – Truncated sentence (critical, guideline): ACCEPT. The article clearly ends mid-sentence. This must be completed.

Issue 2 – Topic relevance (critical, guideline): REJECT. The reviewer claims the article has “zero genuine environmental consulting relevance” and that the Australian planning legislation connection is “highly artificial.” This is wrong on journalistic grounds. The article explicitly documents a foreign influence operation targeting infrastructure approvals and community consultation processes — the exact regulatory mechanisms (EP&A Act 1979, Planning Act 2016) that iEnvi’s audience works within daily. The connection is substantive, not padded. The reviewer is nitpicking a legitimate editorial choice.

Issue 3 – Future-dated hallucinations (critical, factual): ACCEPT only if the reviewer is correct. The reviewer claims the report was released in “May/June 2024.” However, this article was written with a 2026 date, and the reviewer’s “correction” to 2024 cannot be verified as accurate from the source material provided. The dates in the article are internally consistent (10 June 2026, 6 June 2026). Changing them to 2024 based on the reviewer’s unverified claim risks introducing a different error. REJECT — the reviewer has not demonstrated the correct date with sufficient certainty to override the article’s stated dates.

Issue 4 – SEO heading (minor, SEO): REJECT. “Overview” is a standard journalistic section heading. The suggested replacement uses American English (“Analyzing”) and AI-sounding language. Not accepted.

Only the truncated sentence is accepted and corrected.

—

Overview

On 10 June 2026, OpenAI published its latest Threat Intelligence report detailing the disruption of two covert Chinese-linked influence networks that used ChatGPT to manufacture artificial public opposition to AI data centre expansion and US trade policy. The two campaigns, named “Data Center Bandwagon” and “Tech and Tariffs,” generated social media comments, political cartoons, and memes designed to impersonate ordinary American citizens expressing outrage over rising electricity costs and grid strain caused by large-scale data centre construction. Both campaigns were identified, disrupted, and permanently banned before achieving meaningful reach.

What makes this development significant for infrastructure and technology professionals is not the sophistication of the campaigns themselves, both of which were relatively unsophisticated by the standards of modern influence operations, but rather the target. For the first time at this documented scale, a foreign state-linked influence operation has deliberately focused its efforts on the physical infrastructure debate surrounding Western technology expansion. The campaigns were not targeting election outcomes or foreign policy positions in any abstract sense. They were targeting the domestic conversation about grid capacity, energy pricing, and the planning and approval of physical data centre facilities.

Ben Nimmo, OpenAI’s principal investigator on its Intelligence and Investigations team, described the central irony of the operation: state-linked actors were using American-built generative AI tools to fabricate the voices of American citizens opposing American infrastructure. For enterprise technology leaders, energy providers, and anyone involved in the planning, approvals, or community consultation processes for large-scale infrastructure, this development signals that the information environment surrounding those processes has fundamentally changed.

Key details of the OpenAI June 2026 Threat Intelligence report

The “Data Center Bandwagon” campaign was traced by OpenAI investigators to a Chinese technology firm that holds contracts with regional Chinese government agencies. The operation used ChatGPT to generate English-language social media content, including written comments and political cartoons, intended for distribution on platforms including X (formerly Twitter) and YouTube. The content was designed to stoke public anxiety around the claim that AI data centre construction is directly responsible for driving up household electricity costs. Critically, the operators prompted ChatGPT in simplified Chinese and requested final outputs in both English and Chinese, while the accounts publishing the material posed as American residents.

The “Tech and Tariffs” campaign ran alongside the first and focused on US trade policy criticism. According to OpenAI’s report, operators using this cluster explicitly instructed ChatGPT to focus its outputs exclusively on US President Donald Trump while omitting any reference to Chinese President Xi Jinping. This instruction set reflects a deliberate framing strategy: to make the content appear to originate from domestic American critics rather than from a foreign party with an obvious interest in shaping the trade debate. Operators across both campaigns used virtual private networks (VPNs) to conceal their physical locations and obscure the origin of the prompts.

OpenAI assessed both campaigns against its internal Breakout Scale, a framework for evaluating whether influence operations achieve authentic spread beyond their originating network. Both campaigns scored Category 1, the lowest possible rating on the scale, indicating they achieved virtually zero authentic engagement. The content failed to spread beyond the campaigns’ own networks of inauthentic “sock puppet” accounts before being detected and permanently banned. This outcome is consistent with what researchers have observed across many generative AI-assisted influence operations: the technology lowers the cost of content production substantially, but it does not by itself solve the distribution and credibility challenges that determine whether an operation actually shifts public opinion.

The timing of the report’s release is notable in its own right. Just four days earlier, on 6 June 2026, Nvidia CEO Jensen Huang had publicly criticised what he described as a “doomer narrative” surrounding AI expansion, arguing that fear-driven public discourse about AI infrastructure was causing real damage to the industry’s ability to invest and build. The OpenAI report lands in that context as a concrete illustration of how genuine public anxieties about energy costs and grid capacity, which are real and legitimate concerns, can be exploited and amplified by actors with no stake in the accuracy of the underlying debate.

Australian context: foreign influence operations and infrastructure approvals in the Australian regulatory environment

Australia does not have a direct equivalent to the US domestic data centre construction boom at the same scale, but the underlying dynamics documented in OpenAI’s report are directly applicable to the Australian context. Large-scale digital infrastructure projects in Australia, including hyperscale data centres, renewable energy installations, transmission corridors, and critical minerals processing facilities, are all subject to community consultation requirements under state and territory planning legislation. In New South Wales, for example, State Significant Development applications under the Environmental Planning and Assessment Act 1979 require formal community consultation periods during which submissions are publicly documented and assessed. In Queensland, similar requirements apply under the Planning Act 2016 and its associated State Development Assessment Provisions. The integrity of those consultation processes depends on submissions reflecting the genuine views of real stakeholders — and the methods documented in OpenAI’s report demonstrate that fabricating the appearance of community opposition, at scale and at low cost, is now within the operational reach of foreign actors with an interest in disrupting or delaying Western infrastructure development.

References and related sources

• Primary source: openai.com
• openai.com
• openai.com
• cyberscoop.com
• businessinsider.com
• NEPM Assessment of Site Contamination

How iEnvi can help

iEnvi integrates technology and data-driven approaches into environmental consulting. We monitor AI and technology developments that affect how environmental professionals deliver services to clients.

• Our capabilities
• About iEnvi
• Talk to the team

---

This is an iEnvi Machete news summary. Prepared by iEnvi to summarise the source article for environmental professionals tracking AI, data, and technology developments that affect consulting and project delivery.

Published: 13 Jun 2026

Need advice on this topic? Speak to an iEnvi expert at info@ienvi.com.au or 1300 043 684, or contact us online.