As Beijing pushes open models and rapid AI deployment, American policymakers are reaching for regulation and government control.
As Beijing pushes open models and rapid AI deployment, American policymakers are reaching for regulation and government control.
I n an essay published last month, “The Future Is for Everyone,” Meta CEO Mark Zuckerberg rejected the prevailing doom narrative that artificial intelligence destroys jobs and threatens humanity’s core. Transformative progress, he argues, has always faced fear and skepticism, while embracing American values has produced greater prosperity and freedom. The possibility of AI-powered superintelligence in the hands of every American promises to unleash an unprecedented age of discovery, creativity, and enterprise.
Zuckerberg is right. Too many people are running scared when it comes to AI. More troubling, some policymakers seem ready to respond to that anxiety by abandoning the principles that made America a technological and economic powerhouse in the first place.
At first, the biggest concern seemed to be that America would become like Europe: a once-great power in decline, erecting regulation after regulation as though rules themselves create great economies. Europe has spent decades demonstrating that regulatory ambition does not substitute for economic dynamism.
Too many American policymakers still want to follow that path. They look longingly across the Atlantic at the EU’s General Data Protection Regulation and wonder why Americans should not also click through dozens of pop-up notices before accessing a website. They admire the Digital Markets Act and imagine Washington mandating that Apple allow sideloading or dictating how Google displays flights and hotels in search results.
Europe rushed to regulate AI, proposing the AI Act nearly a year and a half before ChatGPT even arrived. The law’s early drafts could not have anticipated foundation models because the technology had not yet emerged as a mainstream policy issue. Yet Europe proceeded to build a regulatory framework around a technology it barely understood.
America’s economic dynamism, especially in the digital economy, has often come from doing the opposite. Some of the country’s most consequential technology policies have removed government barriers rather than added new ones. Section 230 gave internet services the legal foundation to host enormous amounts of user-generated speech. The Internet Tax Freedom Act prevented state and local governments from imposing discriminatory taxes on internet access and electronic commerce. A light regulatory touch gave entrepreneurs room to experiment, compete, and discover what worked.
The bigger problem today goes beyond forgetting those lessons. Some policymakers now seem willing to trade away America’s core principles based on hypothetical and hyperbolic fears about AI.
It started with calls for universal basic income, the brilliant idea that the best way to respond to a technology that makes people more productive is to pay them not to work. After all, the American economy became the world’s largest by making work optional.
Or consider the growing calls for an “FDA for AI,” under which the government would conduct a premarket review before companies release advanced models. Such a regime would end permissionless innovation for AI and give politicians enormous leverage over the technology. A future administration could decide that a model poses unacceptable risks because of its capabilities, its outputs, or the policies of the company that developed it. Each change in political leadership could turn AI regulation into a new battleground over which technologies Americans may use.
That sounds less like the American tradition of technological freedom and more like the governing philosophy of an authoritarian state. Governments that demand ideological conformity from technology companies can easily turn control over technology into control over information.
Then there are the proposed AI “kill switches” — legislative mandates requiring companies to give the U.S. government the technical capability and legal authority to shut down their AI systems.
Frontier AI developers and operators are responsible for the systems they build and deploy and should be accountable for the risks they create. But giving the federal government standing authority to disable privately developed and operated technology is a very different proposition.
That idea should make Americans uncomfortable. The last time a major government contemplated building a kill switch for an entire communications network, it was Russia considering how to disconnect itself from the global internet.
And perhaps the most extraordinary proposals involve government ownership of AI companies, from semiconductor manufacturers to frontier AI labs. OpenAI CEO Sam Altman floated a 5-percent government stake in AI companies. Senator Bernie Sanders (I., Vt.) has called for stakes ten times that size.
What could possibly go wrong? Imagine Anthropic run as efficiently as the Postal Service, or OpenAI developing new models at the speed of Amtrak.
The irony is that China is increasingly pursuing some of the very things that made America the world’s technological leader: open technology, rapid commercialization, massive infrastructure investment, and aggressive deployment. But Beijing is not doing this because it has suddenly discovered the virtues of economic freedom. It is doing it because the Chinese Communist Party (CCP) understands that technological power is economic power, and economic power is geopolitical power.
Beijing wants Chinese AI companies to build globally influential models, embed Chinese technology in foreign markets, and make other countries dependent on Chinese hardware, software, and digital infrastructure. Open models can serve that strategy particularly well. Releasing model weights can accelerate adoption, attract developers, establish technical standards, and make Chinese technology part of the software stack used around the world. It can also allow Chinese firms to benefit from improvements made by researchers and developers elsewhere.
But open does not mean free. The CCP maintains extensive controls over information, requires companies to comply with state directives, and demands ideological conformity. Beijing can encourage openness where it advances Chinese interests while maintaining tight political control.
China’s approach is therefore not laissez-faire capitalism, but state-directed capitalism: using markets, private companies, open technology, and competition when they strengthen national power.
Americans should pay attention to what China is doing not because it should copy China, but because Beijing’s strategy reveals which capabilities will matter in the AI competition.
Who is developing leading open-weight AI models? China.
Who is building the digital and energy infrastructure needed to support AI? China.
Who is accelerating AI deployment in manufacturing and other industries? China.
Who is making it easier for the rest of the world to adopt its AI? China.
Beijing is not embracing open models and rapid deployment because it has embraced American capitalism, but because it believes those capabilities will help China win the AI race.
That should concern American policymakers. If China is racing to build and deploy AI while America constrains its own companies and economic system, the United States risks weakening its greatest strategic advantage: its capacity for innovation.
The choice is not between Chinese state capitalism and European regulation. America has a third option: American capitalism. The lesson is not to copy China’s political system, but to recognize that openness, competition, infrastructure, and rapid deployment can be sources of national power.
The United States will not win the global AI race if its political class becomes more suspicious of capitalism’s creative energy than Beijing’s regime. China is willing to harness private companies, competition, open technology, and market incentives when they advance Chinese power. America should be equally willing to harness those forces — without putting them under state control.
The contrast is becoming particularly stark. Some of America’s leading AI executives are now asking Washington to slow the technology down. Beijing has rejected the proposal, with its state-run media portraying these efforts as an attempt to restrain Chinese AI development and preserve American technological dominance.
Reasonable people can disagree about how quickly frontier AI should advance and what safeguards are necessary. But what happens if America increasingly treats technological progress as something to be constrained while China treats it as something to be won?
The danger is that America could voluntarily give up one of its greatest advantages — its capacity for rapid, market-driven technological innovation — while China continues capturing the economic and geopolitical benefits of AI.
China is not going to handicap itself because American policymakers have become uncomfortable with rapid technological progress. If anything, Beijing has an incentive to welcome policies that slow American innovation.
That does not mean government has no role in addressing AI risks. It has legitimate responsibilities involving national security, consumer protection, intellectual property, and other harms. Many risks can be handled through existing or sectoral rules: medical devices can be subject to medical-safety requirements, autonomous vehicles to transportation and product-safety laws, and AI-enabled fraud or discrimination to laws governing those harms. Genuinely novel risks may require targeted safeguards, but they do not justify a broad licensing regime for AI itself.
For example, if AI makes sophisticated cyberattacks cheaper, the answer should be to strengthen the cybersecurity of critical infrastructure and other vulnerable systems, rather than simply restricting the AI technology that lowers the cost of hacking. The U.S. government also has an important role in creating the conditions for technological progress with policies that encourage energy production, computing infrastructure, scientific research, and AI adoption. The goal should be to address specific risks while leaving the underlying technology as free as possible to develop.
The nation faces a formidable competitor in China. To prevail, the United States needs to draw on the principles that made the American model successful in the first place: economic freedom, technological ambition, and competition.
America’s greatest technological advantage has never been the government’s ability to predict which technologies will succeed. It has been the ability of millions of entrepreneurs, investors, researchers, workers, and consumers to discover what works through competition and experimentation.
AI could usher in an extraordinary era of abundance and human achievement. America should compete for that future by remaining unmistakably American — by being free, ambitious, and confident enough to let its people build the future.