Welcome to AI Arms Race, a quarterly newsletter unpacking the ends, ways, and means of the geostrategic war over critical technologies. I’m Courtney Manning, and I lead AI Imperative 2030—an ASP program driving high-impact research on the threats posed by China’s aggressive pursuit and weaponization of artificial intelligence.

Last week, I published a report on how industry and trade interventions impact American innovation, which you can read here. Bottom line up front: The United States can't regulate its way to global AI dominance, but we could certainly regulate ourselves out of it.

Only incentives drive innovation; industry controls are deterrents and guardrails. That’s not to say AI regulations are unnecessary or that the private sector deserves a blank check. On the contrary, excessive subsidies and tax breaks distort markets, fuel protectionism, and hinder innovation dynamism. That’s why U.S. industrial controls must be motivated by true national security imperatives, not economic growth aspirations.‍

But what are those “true” national security imperatives, and how can policymakers ensure that the ends justify their means?

‍First, “strategic competition” may be an economic moniker, but the U.S. and China are not just in a tech race—we’re in an arms race. Artificial intelligence will redefine how we live and work, but also how we wage war and preserve peace. If Beijing’s computing power, energy supplies, AI hardware and software, and weapons systems surpass those of the United States, the Chinese Communist Party could overcome U.S. and allied defenses, compromise U.S. critical infrastructure, and shift the global balance of power from a democratic rules-based order to isolationism and authoritarianism.

Beyond these grand strategic threats, near-term risks are mounting. Chinese firms lag far behind their American counterparts in AI safety, demonstrating little to no regard for the technical safeguards necessary to prevent catastrophic failures. The same is true for data privacy and intellectual property rights; TikTok faces over 1,500 data privacy lawsuits, while thousands of terabytes of user data from WeChat and DeepSeek are being funneled into China’s surveillance algorithms.

These efforts are central to Beijing’s asymmetric tradecraft strategy to “leapfrog” the United States, justified by CCP propaganda as “reparations” for Western subjugation beginning in China’s century of humiliation. But there is no point at which this debt will be considered repaid or fairness restored.

Nor is the CCP concerned with its own citizens’ safety as it aggressively pursues these objectives. China has no equivalent to the AI Safety Consortium or U.S. AI Safety Institute, nor any legitimate avenue for public debate on AI governance. Unlike in Washington, Chinese officials face no accountability for atrocities that may result from a lack of risk management and cannot be democratically replaced to prevent such failures from recurring.

For these reasons and more, American leadership in artificial intelligence is a national security imperative. It’s not just a race, but a battle of opposing forces. On one front stands visionaries, humanitarians, whistleblowers, and ethicists; on the other, functionaries in the CCP’s civil-military fusion regime. But principles alone cannot deter China’s whole-of government pursuit of weaponized AI. If we fail to equip our most ethical and capable innovators with the tools they need to fight fair, we risk yielding our future to those who do not believe in restraint.