Americans Benefit From WTO Rules Governing the Free Flow of Data

Lori Wallach and Daniel Rangel wrote a thoughtful analysis in Tech Policy Press about the 2023 change in the US position on digital trade. They argued that “[s]ome of these proposals would set binding and enforceable constraints on domestic tech policy, affecting online privacy and civil rights, competition rules, AI oversight, and even prospective gig worker policies.” Hence, they conclude that the Biden Administration was right to withdraw its support for these positions, implying that it is necessary to retain policy space to effectively regulate these areas before committing to international agreements that could restrict such efforts.

I respectfully disagree, for two reasons:

First, I have learned from 13 years of research on digital trade and data governance that we can simultaneously push for regulations at home while negotiating shared rules abroad. Moreover, under existing trade rules, the US already has that policy space under trade rule exceptions, which recognize that policymakers must be able to breach international rules to achieve equally important domestic policy goals such as protecting privacy and regulating the data giants.

Secondly, the world is connected by data. Those connections, whether through Facebook or research on COVID, have kept us sane, social, and informed during the pandemic, enabled cross-border research on vaccines, and allowed us to continue learning and working. But those data flows need to be matched and governed consistently with internationally accepted rules. Whether we like it or not, the WTO is where some 90 governments are negotiating those rules.

I’ve been writing on digital trade since 2011, when I received grants from the Ford and MacArthur foundations to examine the human rights impact of cross-border data flows. I learned that the free flow of data can both advance and undermine human rights such as freedom of speech and association, and good governance. I also found that data-driven technologies can be used by repressive states, terrorists, and complicit firms to undermine human rights and advance despicable business practices, such as what’s now called surveillance capitalism. These are both domestic and international problems that require both domestic and international mitigating strategies and rules. 

The WTO sets rules governing how nations can trade and how and when they can enact regulations that may discriminate between domestic and foreign market actors. Currently, 90 nations are participating in these talks, called the Joint Statement Initiative on E-Commerce. These countries have divergent views on how best to nurture the digital economy and what role trade agreements should play in governing it. They also have different levels of digital prowess and the ability to participate effectively in these talks. But they all understand this: without data, they cannot trade, without trade, their economies will grow less or not grow at all. These officials comprehend that they must shape the negotiations. US officials should, too–here are a few reasons why.

First, most economies are built on a foundation of data. Where nations once grew rich solely from their natural resources and human capital, nations today also rely on their citizens’ ability to collect, analyze, and create goods and services built on large pools of data. But data is multidimensional—the same dataset can be both a commercial asset and a public good, which governments should provide for and regulate effectively. Much of those datasets are global, such as those that are scraped from the web that underpin AI chatbots such as ChatGPT.

Second, support for the free flow of data is not just integral to trade policy–it is also essential to the open internet and, thus, America’s support for democracy around the world. The US-led Declaration on the Future of the Internet commits signatories to “[p]romote our work to realize the benefits of data free flows with trust based on our shared values as like-minded, democratic, open and outward looking partners.” In its 2022 National Security Strategy, the Biden White House asserted that the US is “rallying like-minded actors to advance an international technology ecosystem that … promotes the free flow of data and ideas with trust, while protecting our security, privacy, and human rights, and enhancing our competitiveness.”

Third, support for the free flow of data through trade agreements is a longstanding US policy. In 1997, then President Bill Clinton announced the Framework for Global E-Commerce in recognition that the world needs rules to govern data flows. It states, ‘The US government supports the broadest possible free flow of information across international borders [. . .] The Administration [. . .] will develop an informal dialogue with key trading partners [. . .] to ensure that differences in national regulation [. . .] do not serve as disguised trade barriers.” Policymakers were essentially arguing that nations would regulate differently at different times—we would have to accept that, but that we can accommodate these differences through rules plus clear exceptions to those rules that acknowledge the need to maintain domestic policy space (as an example to protect privacy or national security.)

These policies have been encapsulated in bilateral and regional trade agreements that enshrine the free flow of data/information (whether scientific information, gossip, or news) as well as language limiting how and when other nations can erect barriers to digital trade. For example, the Japan-US Commercial and Industrial Partnership, issued on November 14, 2023, notes, “we will continue collaborating to facilitate cross-border data flows and effective data and privacy protections globally, in support of our efforts to enable cross-border data flows and operationalize Data Free Flow with Trust.” The US has also drafted and supported language at the most recent G7 and G20 meetings–and for good economic reason: Digital trade underpins traditional trade in goods and services, but digital trade is growing faster than traditional trade. Hence, even under the Trump Administration, the US led efforts at the WTO to set rules governing digital trade. See as an example the US white paper of 2019, which gives a wide range of economic, social, political, and human rights reasons why the free flow of data benefits the world’s people.

But nations are adopting policies that alter market conditions and distort trade in ways that favor domestic producers of data-driven services over foreign competitors. The OECD Services Trade Restrictiveness Index (STRI) provides annually updated, comparable information on regulations affecting trade in services across 50 countries and 22 sectors from 2014-24. This metric of barriers to cross-border computer services can serve as a proxy. The OECD defines computer services as computer programming, consultancy, and related activities, and information service activities such as artificial intelligence-based platforms including ChatGPT, apps such as Strava, and applications such as Google Maps. The OECD found a gradual increase in these barriers. These statistics should tell us that we need clearer rules in the interest of data subjects (individuals such as you and I) and data service providers such as TikTok, the New York Times, or the US Department of Commerce.

Yet in late October 2023, trade journalists reported that “the Biden administration will end its support for proposals on data flows, data localization and source code being discussed in the World Trade Organization.” US officials stated that they need “policy space.” The news surprised trade diplomats, researchers, human rights activists, and longstanding trade observers. It was a major break with US policy.

To this author, the arguments made by USTR (and proponents on Capitol Hill) did not hold water. First, the US would not in any way lose the ability to preserve its policy space. Every trade agreement contains exceptions, designed to ensure that policymakers can meet key domestic responsibilities should achieving those responsibilities clash with domestic objectives. Hence, nations can breach the rules to protect users from harm, preserve social stability, and protect national security. Secondly, these rules do not just benefit big tech–they benefit tech users, by providing certainty regarding data flows.

Moreover, the language of ‘policy space’ was reminiscent of what the world’s most populous country, India, said when it refused to participate in the global trade talks on digital trade. India said it would not join negotiations on the free flow of data because it was sovereign over its people and firm’s data. But that argument is essentially built on the notion that India’s data is different from that of other countries because it has only one purpose: to benefit the nation and people of India. Such a perspective ignores the very real benefits of data—the same set of data can be used and reused by many different entities in many countries over time. We should not undermine the potential of insights from data to help advance human rights, accountability, and to mitigate wicked problems, such as climate change.

Secondly, the announcement was inconsistent with recent US policy pronouncements at the G-7, the OECD, and even at America’s own Democracy Summit. Unfortunately, the US announcement signaled to America’s trade partners that the US was no longer committed to ensuring internet openness and open access to information. The US has long pushed for greater international access to a wide range of data sets in the belief that if personal data is protected, datasets can be reused in different ways for different purposes.

Support for the free flow of data is not just integral to trade policy–it is also essential to the open internet and America’s support for democracy around the world. The US-led Declaration on the Future of the Internet commits signatories to “[p]romote our work to realize the benefits of data free flows with trust based on our shared values as like-minded, democratic, open and outward looking partners.” In its 2022 National Security Strategy, the Biden White House asserted that the US is “rallying like-minded actors to advance an international technology ecosystem that … promotes the free flow of data and ideas with trust, while protecting our security, privacy, and human rights, and enhancing our competitiveness.”

Thirdly, we should want to create the most accurate, complete, and representative datasets possible so that the AI that US developers produce is trustworthy, reliable, and as unbiased as possible. The US National Institute of Standards and Technology (NIST) has warned that many variants of AI depend on “large scale datasets, which can lead to data quality and validity concerns. The difficulty of finding the ‘right’ data may lead AI actors to select datasets based more on accessibility and availability than on suitability… Such decisions could contribute to an environment where the data used in processes is not fully representative of the populations or phenomena that are being modeled, introducing downstream risks.” In short, NIST is concerned with problems of both quality and validity. However, by relying on data scraped from the web, LLMs are likely producing incomplete and inaccurate outputs. Scraped data, in essence, provides a snapshot of the internet in time, but it is likely an incomplete, incorrect, outdated picture.

However, one can only scrape the world wide web that exists, not the web we wish to see. The web is dominated by content from and about people who are online, and those people live mainly in Europe, North America, and Asia. Throughout Europe, the Commonwealth of Independent States (CIS) and the Americas, between 80 and 90 percent of the population uses the Internet, approaching universal use (defined for practical purposes as an Internet penetration rate of at least 95 percent). Approximately two-thirds of the population in the Arab States and Asia-Pacific countries (70 and 64 percent, respectively) use the Internet, in line with the global average, while the average for Africa is just 40 percent of the population. However, in 2022, the ITU reported that 34% of the world’s population had never used the internet. Most of these people live in rural areas in the developing world. Data related to these people are not visible in most web scraping. But companies are under pressure to make their datasets more accurate, representative, and complete. They can do so with synthetic data, making up datasets that are proprietary or inferred from broader datasets. But broader datasets are only possible if AI developers have greater access to more of the world’s data, something that is only possible through a multilateral agreement that sets clear rules governing data with exceptions.

Finally, the regulation of data flows is fragmented and incomplete. The EU, the US, Australia, and other countries have now negotiated a wide range of digital trade and e-commerce agreements with different provisions. I believe that Americans will benefit if the rules are less fragmented, more up-to-date, and clear about how various types of data can be regulated. Americans need to be in the room where negotiations about data happen. At the same time, we must continue to pressure Congress to regulate the data giants, to make their datasets more accurate, complete, and representative, and to set rules governing personal data.