Sources: the US has found it difficult to distinguish between tech that China can use for its military and tech used commercially by firms, particularly with AI
Defining risky artificial intelligence poses challenge as Washington moves to curb investment in Chinese tech Tweets: @davidmwessel , @bobdavis187 , and @jchengwsj Tweets: David Wessel / @davidmwessel : About those outward FDI rules: Distinguishing between technology that Beijing could use to advance its military and technology that Chinese companies use for everyday commercial purposes has proven difficult, particularly when it comes to AI https://www.wsj.com/... via @WSJ Bob Davis / @bobdavis187 : This is one reason for continued engagement. AI regulation needs to be global to be effective and fair. China needs to be a bit part of those discussions. https://twitter.com/... Jonathan Cheng / @jchengwsj : Overheard in Washington: “There's never going to be any investment in China that doesn't pose any risk...What the administration needs to decide is how much risk they can take on and draw the line there.” @aduehren @ryanjtracy https://www.wsj.com/...
Context & Ripple Effects
The definitional problem at the heart of this story has been visible since the first round of AI controls: when the Commerce Department floated AI export restrictions in early 2019, industry insiders warned the rules would stunt the US industry itself, and the Trump White House spent that year debating how hard a line to take on China's access to American tech. The WSJ's David Wessel, Bob Davis, and Jonathan Cheng now report the same ambiguity blocking a newer instrument — outbound FDI rules meant to stop US capital from funding tech Beijing could militarize.
What makes the piece consequential is that neither side is converging on precision. By 2025, Trump officials were weighing whether to scrap the Biden-era three-tier AI chip rule altogether, while Beijing had instructed its top AI leaders to avoid visiting the US over detention fears — signs that both governments treat AI's blurry military-commercial boundary as something to route around rather than define.
First-order effects
- Drafters of Washington's outbound investment rules face a binary with no clean answer: define 'risky AI' narrowly and dual-use systems slip through, or broadly and US capital flowing to ordinary commercial Chinese AI gets caught in the net.
- US investors in Chinese AI startups and the Chinese firms seeking that capital face deal-by-deal scrutiny against criteria that, per the reporting, even the agencies writing the rules cannot pin down.
Second-order effects
- Compliance uncertainty pushes Chinese AI fundraising toward the domestic channels Beijing is already building — state media reported three new venture funds of over $7.1 billion each aimed at early-stage hard-tech startups — eroding the curb's leverage before it takes effect.
- A rule that cannot distinguish targets invites overcorrection: US firms engaged in legitimate commercial AI work with Chinese partners bear screening costs and deal delays regardless of whether their activity is ever deemed sensitive.
Third-order effects
- If the pattern holds, controls migrate from technology definitions to blunter structural instruments — country tiers, entity designations, outright ownership bans — precisely because the dual-use line in AI keeps resisting being drawn.
- The blur frays people-to-people ties symmetrically: Beijing treating US travel by its own AI researchers as a security risk extends decoupling from capital flows into scientific collaboration itself.
The trend: US-China tech policy is drifting from targeted, definition-based controls toward blunter geographic and structural ones, because AI's military-commercial boundary has proven impossible to legislate cleanly.