Dragonbridge: the PRC-linked network now targeting Imperial Valley

What Dragonbridge is

Threat researchers first identified the network in 2019. The initial activity was relatively crude: low-quality, pro-Beijing narratives pushed by inauthentic accounts that drew minimal engagement. Six years later, the network has evolved dramatically. Google's Threat Analysis Group has reported disrupting tens of thousands of Dragonbridge channels and accounts across YouTube, Blogger, and Gmail. Meta has named it as one of the largest Coordinated Inauthentic Behavior takedowns in the company's history.

The network's strategic role is straightforward. It is one of the People's Republic of China's primary tools for shaping the global information environment around technology competition, critical minerals, and U.S. domestic politics. The operations are not designed to convince readers of any specific argument. They are designed to amplify hostility, sow distrust, and consume the attention of municipal decision-makers.

From Spamouflage to MAGAflage

The Institute for Strategic Dialogue documented a sharp tactical shift starting in 2023. Operatives stopped trying to seed pro-PRC content directly and instead began impersonating culturally specific U.S. demographics. Researchers nicknamed the new tactic MAGAflage because some operatives adopted right-wing American personas, complete with U.S. flag imagery, regional slang, and patriotic framing.

Other operatives in the same network adopt the opposite profile: deeply concerned local environmental activists in communities that already have legitimate environmental grievances. The targeting is precise. The personas are tuned to whatever cultural frame is most likely to amplify division in a given local debate.

Recent campaigns make heavy use of generative AI. Threat researchers have documented Dragonbridge using AI to generate large volumes of text with fewer linguistic errors than human operators could achieve alone, summarize trending social media posts to identify wedge issues, and even produce deepfake video anchors for fictitious news outlets. This is what makes the current threat materially different from 2019.

The Texas rare-earth precedent

The strategic logic is publicly documented: roughly 80 percent of U.S. lithium reserves lie within 35 miles of Native American reservations or marginalized communities that have legitimate, long-standing environmental justice concerns. Foreign operators do not need to invent a grievance. They need only identify an existing one and pour digital fuel on it until the local development becomes politically impossible.

Why Imperial Valley is the priority target

Imperial County uniquely combines two strategic prizes:

The Salton Sea region — colloquially called Lithium Valley — sits directly adjacent to the proposed data center site. Securing this lithium supply is an absolute strategic imperative for the United States to achieve independence in electric vehicle batteries, grid-scale energy storage, and the advanced electronics required for modern data centers themselves.

The PRC has explicitly designated AI infrastructure as a national security priority and launched a state-coordinated buildout called Eastern Data, Western Computing (Dongshu Xisuan), which migrates energy-intensive computing from the densely populated eastern seaboard to resource-rich western provinces. China's 2021–2023 Data Center Plan set a target of 20 percent annual growth in server rack installations, supported by an estimated $1.4 to $2.8 trillion in tech-sector investment this decade.

Against that backdrop, slowing the Imperial Valley project is not a localized squabble. It is a tactically meaningful move in a global infrastructure race.

What platforms have publicly disrupted

• Google Threat Analysis Group has publicly disclosed disrupting tens of thousands of Dragonbridge accounts across Google services, including a major Q1 2024 enforcement action.
• Meta has removed Dragonbridge clusters as part of its broader Coordinated Inauthentic Behavior enforcement — more than 200 networks across 68 countries since 2017.
• Canadian government researchers (Rapid Response Mechanism Canada) have documented Dragonbridge operations targeting Canadian politicians as part of broader transnational repression campaigns.
• OpenAI has reported disrupting covert influence operations using its tools, including networks aligned with the same broad strategic objectives as Dragonbridge.

These disclosures matter for one reason: they confirm the network is real, ongoing, and that platforms themselves can identify and remove activity at scale. But platforms also publicly acknowledge that automated removal cannot keep pace with the volume and sophistication of these campaigns. Public awareness is part of the defense.

What actually helps

Threat researchers consistently emphasize a few defenses that are unusually effective against this kind of operation:

See also

Outside Influence Watch — the full overview, including how to spot CIB →