U.S. Weighs Human Rights As It Looks To Expand China Chip Restrictions

Ahead of Treasury Secretary Janet Yellen’s ongoing trip to Beijing, The Wall Street Journal reported the Biden Administration is considering new restrictions on advanced semiconductor exports to China. They would prevent U.S. firms, including Nvidia, a company that designed special chips meant to conform to the October 2022 export controls, from selling advanced chips to China.

These moves are depicted as protecting the United States’ national security interests and competitive technological and economic edge against China. But another, less public, impetus for the curbs is a factor that has long been central to the U.S.-China relationship: human rights.

Along with like-minded partners, the Biden Administration is pursuing a strategy of “advancing technology for democracy.” This initiative is manifested in multilateral documents like the Declaration for the Future of the Internet, the AI Bill of Rights and other tech-focused agreements that come out of the now-annual Summit for Democracy.

The purview of “democratic tech” has expanded to explicitly include export controls, evidenced in the Export Controls and Human Rights Initiative, which was established at the first summit. According to this year’s readout, that initiative aims to “integrate a human rights lens into export controls so as to prevent dual-use technologies and goods from falling into the hands of those that may misuse them.” It also introduced a Code of Conduct for Enhancing Export Controls of Goods and Technology. It calls for “subscribing states” (other nations, presumably democracies, that opt in) to “take human rights into account when reviewing potential exports of dual-use goods, software or technologies.” Twenty-four countries have signed on.

It makes sense that the human rights rationale for export controls is aimed abroad, since plenty of countries are more interested in signaling their support for human rights than partaking in pervasive U.S.-China competition. (Domestically, however, the competition angle plays very well; a 2023 Pew survey found that 83% of Americans have a negative view of China and four in 10 Americans consider China an “enemy.”)

Administration officials would say the restrictions are part of a broader strategy to de-risk, not decouple. The term “de-risk” was borrowed from European Union Commissioner Ursula von der Lyen, who uses it to espouse a tough-on-China approach that doesn’t cut off economic ties. U.S. lawmakers, who are publicly and objectively more hawkish on China, are likely using it to signal a less existentially fraught China policy that would ideally placate Brussels, if not Beijing.

One read of the language coming out of the Administration suggests the difference between de-risking and “complete decoupling”—which is favored by some GOP presidential candidates like Vivek Ramaswamy—lies in scope. Per some officials, the controls are limited to securing America’s military advantage, and do not extend to stifling China’s economic growth. As national security advisor Jake Sullivan put it, the measures are part of a “small yard, high fence” approach wherein only the most crucial technologies are resolutely defended. According to Sullivan, the restrictions are “simply ensuring that U.S. and allied technology is not used against us.”

The degree to which the U.S. export regime is genuinely that focused on military limitations is up for debate. Since the controls affect other industries that rely on the restricted chips—most notably China’s AI industry—Beijing’s claims that the moves are an effort to undercut China’s technological and economic development have significant supporting evidence, most notably from lawmakers in Washington who themselves explicitly state the competition with China as a motivation for securing the United States’ continued technological leadership.

Most significantly, competition with China is used to argue against U.S. regulations on artificial intelligence, such as the framework introduced by Senator Chuck Schumer last spring. As of now, the United States has no regulation on AI applications ranging from chatbots to facial recognition-enabled surveillance. The “small yard, high fence” national security framework and highbrow moralism of the Export Controls and Human Rights Initiative is matched with minimal protections for U.S. citizens. The human rights argument falls flat in this context.

The United States should design regulations that protect human rights in end use applications of AI. But as of now, the non-binding statements that comprise “democratic tech” remain virtue signaling that fails to earnestly protect any specific virtue. Eight months after the release of ChatGPT, OpenAI’s toughest opposition has come not from government, but in the form of a class action lawsuit out of California that alleges the company violated innumerable people’s copyright and privacy protections when it processed nearly the entirety of the internet to train its model.

OpenAI reportedly lobbied lawmakers in the EU to pare down the AI Act when it was in draft stage in 2022. Part of the company’s reasoning was that they had taken all steps necessary to mitigate against nefarious or “high risk” uses of AI internally. There was no need for external checks. The EU went ahead with its AI Act anyway. The U.S. still has nothing close.

This laissez-faire domestic approach contrasts harshly with the active stance the U.S. takes to prevent China’s technological development, national security threats and/or human rights abuses. When China is invoked, the U.S. has proven itself willing to incrementally and increasingly curb hardware access, taking effectual, expensive steps to do so—including some that harm the interests of American businesses like Nvidia.

There are justifications for export controls to the PRC. Specifically, there are credible arguments for restricting China’s access to the most advanced chips. But countries like those in the EU, which have actually come up with comprehensive regulations on artificial intelligence, have much sturdier ground to stand on than the United States in condemning China’s uses of such technologies.