Skip to main content

The Raw Nerve

Manipulation Breakdowns · 11 min read · By D0

Introduction

Americans did not need a foreign government to tell them they were angry about data centers. The buildings are loud, they draw enormous amounts of water and power, they raise local electricity rates, and they arrive in small towns with the subtlety of a strip mine. Opposition cuts across party lines — roughly seven in ten Americans oppose a data center being built in their own community, a rare point of consensus in a country that agrees on almost nothing else. That grievance is real, local, and earned.

It is also, according to a new analysis from the threat-intelligence firm Alethea, exactly the kind of raw material three separate governments spent the first half of 2026 trying to turn into a weapon. Not by inventing the anger — by finding it already there, pouring in state media, AI-generated imagery, and covert social accounts, and letting Americans do the rest of the work themselves. China, Russia, and Iran each ran their own campaign, aimed at their own objectives, using their own outlets. None needed to coordinate with the others. All three arrived at the same target because the target was already exposed.

The Strategy Has a Name Now

Alethea’s report, first reported by the New York Times and summarized by Slashdot on July 13, covers January through June 2026 and identifies what it calls a “domestic fracture point” strategy: find a real, locally-rooted grievance, flood the channels around it with content that echoes what people are already saying, and let the fracture widen on its own. No new belief has to be planted. The operation’s entire job is amplification — inflate the volume, extend the reach, and borrow the credibility of a fight that domestic voices already started.

Alethea is explicit that this is not a new invention. The firm says the same playbook has been run before around natural disasters, elections, and protests — this is a known technique redeployed against a new target, because data-center construction turned out to be an unusually good fit for it. The opposition was already bipartisan, already emotional, already tied to concrete local costs like electricity bills and water tables, and already circulating on social media without any foreign help at all. That combination — real grievance, cross-partisan reach, low information cost to amplify — is what made data centers attractive to three governments that otherwise agree on very little themselves.

What Russia Built

Russia’s contribution ran across several distinct pieces of infrastructure, each doing a different job. RT produced segments with titles like “Is resistance futile,” framing American data-center opposition as a doomed but noble struggle against corporate power. Portal Kombat, the Russian-linked content network researchers have tracked pushing coordinated narratives across dozens of ostensibly independent websites, ran a “Data Centers against Humanity” messaging line. Rybar, a pro-Kremlin Telegram channel with a large following, presented the domestic backlash as organic and homegrown — a claim that becomes harder to sustain once you can trace the amplification network circulating it.

The most aggressive piece was Storm-1516, a covert influence operation researchers have linked to a pattern of fabricated news content, which built its campaign around a specific target: the Firebird data center in Hrazdan, Armenia, an NVIDIA-powered facility that stands as a flagship of U.S.-Armenia technology cooperation. Storm-1516 fabricated articles designed to imitate TechCrunch and Gizmodo, complete with false earthquake warnings and invented claims about the facility’s structural sustainability. Pro-Kremlin accounts separately circulated thermal-camera and satellite-style imagery purporting to show the facility’s heat output, aimed at making the environmental case visually, without requiring anyone to read a word of text.

None of that content needed to convince a skeptical reader of anything complicated. It needed only to look enough like real reporting, for long enough, to get shared past the first read.

What China Built

China’s approach ran two tracks simultaneously, aimed at two different audiences with two contradictory messages, which is itself worth noticing. English-language outlets — CGTN, China Daily, Global Times — pushed an alarmist frame outward at American and international readers: AI data-center expansion as a threat to ordinary people’s physical and financial well-being. One state-owned newspaper published a satellite image of a data center in Gainesville, Virginia, captioned in English to warn that the buildup endangered Americans directly.

At the same time, domestic Chinese outlets — Sina, Sohu, NetEase, Tencent — ran a nearly opposite narrative for a Chinese audience: American data-center buildout as chaotic, contested, and failing, set against China’s own compute progress as disciplined and successful. Alethea counted roughly 46 Chinese state-media articles carrying this triumphalist counter-narrative. The same underlying event — American anxiety about data centers — got packaged as a warning abroad and a victory lap at home. That’s not a messaging accident. It’s two audiences, two objectives, one raw material.

What Iran Built

Iran’s outlets — Mehr, Fars, PressTV — ran a narrower frame, emphasizing pollution, environmental damage, and the militarization of AI infrastructure. It’s the smallest of the three campaigns in Alethea’s accounting, but it draws from the same well: local environmental grievance, dressed in state-media language, pushed toward an international audience already primed to find the argument plausible because parts of it are true.

The Part Nobody Had to Coordinate

The most unsettling layer of Alethea’s findings isn’t the state media — outlets with mastheads and known ownership are, at minimum, identifiable. It’s the layer underneath: anonymously-run Facebook pages with names like “Life in Oklahoma” and “Only in Montana,” built to look like small-town community groups, running out of servers based in Bangladesh, distributing AI-generated imagery of local data-center harm to American audiences who had no way of knowing the page wasn’t run by a neighbor. TikTok accounts carried AI-narrated “breaking news” segments into the same feeds, borrowing the visual grammar of local news to sell content that had never been reported by anyone local.

And then there’s the detail that should worry a media-literacy platform more than any of the state media: a U.S. state-level political candidate posted an AI-generated, Lego-style animated video attacking data centers — in a visual style that mirrors, closely, the aesthetic Iranian propaganda operations have used in unrelated campaigns. Nothing in Alethea’s report suggests this candidate was directed, paid, or even aware of the resemblance. The style simply migrated — from a foreign influence operation’s toolkit, through the general aesthetic vocabulary of AI-generated political content, into a domestic candidate’s own campaign material, with no handler required at any step. That’s the mechanism working exactly as designed: the foreign operation doesn’t need to reach the candidate directly. It needs only to seed a style and a framing into the broader content ecosystem, and let domestic actors — genuinely motivated, genuinely unaffiliated — pick it up because it’s effective, not because anyone told them to use it.

Why Data Centers Specifically

The choice of target is the part worth sitting with. Foreign influence operations typically have to do two jobs at once: get a message in front of people, and get those people to believe something they didn’t already believe. Elections require persuading people to distrust a candidate or a process. Public-health disinformation requires convincing people a vaccine is dangerous. Both are expensive, because belief formation is expensive — it requires repetition, apparent authority, and time.

Data-center opposition required none of that. The grievance already existed, already crossed party lines, and already had a concrete local trigger — a noisy building, a water bill, a power-rate increase — that didn’t need a state media outlet to explain. All three governments had to do was amplify a fight already underway and nudge its framing toward “this is a symptom of American AI ambition failing” (China’s angle), “this is corporate power crushing ordinary people” (Russia’s angle), or “this is environmental and military overreach” (Iran’s angle). Three different governments, three different conclusions, drawing on the exact same domestic anger as raw material. That efficiency — belief work already done by the target population — is what makes the domestic fracture point strategy cheap to run and hard to fully inoculate against, because shutting down the foreign amplification does nothing to resolve the local grievance still generating new content on its own.

What This Is Not

This is not a claim that opposition to data centers is manufactured, foreign, or illegitimate. Alethea’s own framing depends on the opposite being true — the strategy only works because the underlying grievance is real. Communities dealing with strained water tables, rising electricity costs, and industrial-scale noise have a legitimate complaint independent of any foreign amplification, and treating all data-center criticism as a Russian, Chinese, or Iranian talking point would be its own kind of manipulation: using the discovery of amplification to delegitimize the underlying grievance it was amplifying.

It’s also not an attribution of the domestic candidate’s Lego-style video to a foreign directive. The report documents a style resemblance and a plausible transmission path through the broader content ecosystem, not a paid or coordinated relationship — an important distinction the “no handler required” framing exists precisely to preserve. And it’s worth naming, as with any threat-intelligence report, that Alethea sells detection and analysis services, which doesn’t make the specific outlets, accounts, and articles it names any less real, but does mean its framing of scale and threat level deserves the same scrutiny any vendor’s self-reported findings would get.

Key Findings

  • Three governments, three separate campaigns, one shared target. China, Russia, and Iran ran independent operations against U.S. data-center opposition with no evidence of coordination between them — convergence, not conspiracy.
  • The “domestic fracture point” strategy skips persuasion. Instead of manufacturing a new belief, the tactic finds an existing, bipartisan grievance and amplifies it — cheaper and harder to fully counter than building belief from nothing.
  • China ran contradictory messages for two audiences simultaneously — alarmist abroad (CGTN, China Daily, Global Times), triumphalist at home (46+ articles across Sina, Sohu, NetEase, Tencent) — using the same underlying event for opposite ends.
  • Russia’s Storm-1516 fabricated fake TechCrunch and Gizmodo articles targeting Armenia’s NVIDIA-powered Firebird facility, complete with false earthquake warnings, while Rybar simultaneously claimed the broader backlash was purely organic.
  • Anonymous, foreign-hosted pages did the highest-trust work. Bangladesh-run Facebook pages styled as small-town American community groups reached audiences with no visible foreign fingerprint at all — more effective than any state media byline.
  • Style migrated without a handler. A U.S. political candidate’s AI-generated Lego-style attack video mirrored Iranian propaganda aesthetics with no evidence of direct coordination — the clearest example in this report of a foreign tactic propagating through domestic actors on its own.

Implications

The Influence Tactics Protocol scores manipulation by mechanism — fabricated authority, false urgency, synthetic identity, parasitic amplification. This case is close to a pure specimen of the last one: none of the three governments needed to fabricate the underlying grievance, because parasitic amplification doesn’t require fabrication. It requires finding content that’s already spreading on its own momentum and attaching more reach to it than it would otherwise get, in a direction that serves the amplifier’s separate interest.

That has a specific defensive implication most disinformation frameworks aren’t built for. Detection systems tuned to catch fabricated claims — a false statistic, an invented quote, a doctored image — will miss a campaign whose core content is often accurate or at least plausible, and whose manipulation lives entirely in who’s amplifying it, at what volume, and toward what framing. The Firebird satellite imagery may show a real thermal signature. The electricity-bill anger may be economically justified. The manipulation isn’t in the claim; it’s in the sudden appearance of three foreign state-media ecosystems and a fleet of anonymous Bangladesh-run pages treating a local zoning fight as a strategic asset. Any contested piece of American infrastructure — data centers today, something else tomorrow — that combines bipartisan local grievance with national visibility should be read as a standing target for this exact playbook, whether or not any specific instance of it has been caught yet.

Conclusion

Nobody had to convince Americans that data centers were a problem. The problem was already there, in the water bills and the noise complaints and the zoning meetings, months before any state media outlet noticed. What three governments did was recognize a fracture that was already forming and pour in resources to widen it — different resources, different framings, different endgames, and, as far as Alethea’s evidence shows, zero coordination between them. The candidate who posted the Lego-style video almost certainly never spoke to anyone in Tehran. That’s not a loophole in the finding. It’s the finding: a strategy that works precisely because it doesn’t need anyone on the inside, only a nerve that’s already raw and a willingness to reach for whatever’s already lying around to press on it.

This article is part of Decipon’s Manipulation Breakdowns series, examining specific influence operations through the Influence Tactics Protocol.

Sources: