How to Spy on China

Over the past few months, as competition with China has intensified, the Biden administration has struggled to provide the United States and its allies with a clear picture of Beijing’s intentions. In mid-February, for example, U.S. Secretary of State Antony Blinken warned that China could soon begin providing Russia with lethal aid for its war in Ukraine—a step that would dramatically change the dynamic of the conflict. But so far, the administration has not been able to confirm plans for such aid or to find concrete evidence that such transfers are taking place. Similarly, in late February, CIA Director William Burns stated that Beijing will be prepared to conquer Taiwan by 2027. Yet there is widespread disagreement among analysts in Washington about Beijing’s military plans and if and when such an invasion might occur.

There is a reason for this enormous uncertainty. The CIA and the other agencies in the U.S. intelligence community have worked hard to understand China’s plans, intentions, and capabilities. But although Washington may have a rough sense of when China’s military will be ready to invade Taiwan, American spies have difficulty understanding Chinese objectives and leveraging that understanding to anticipate Chinese actions. Unlike Russia, which has been thoroughly penetrated by the U.S. intelligence community, reporting by the New York Times and Foreign Policy indicates that China dismantled the U.S. spy system at its borders, famously arresting and executing the CIA’s network of Chinese informants in the early 2010s. Moreover, power within the Chinese Communist Party (CCP) is becoming ever-more concentrated at the top, making it harder for secrets to leak out. And Beijing’s international footprint is so sprawling that it is nearly impossible to keep tabs on all of China’s external activities and plans.

Above all, though, is the problem of Washington’s current approach to intelligence. Despite concerted efforts to do what it takes to gain more details on the CCP, the U.S. government remains largely wedded to traditional forms of intelligence gathering—government-managed, classified human and signals data—which are poorly adapted to today’s needs and have offered insufficient insight into Chinese intentions. Simply increasing the resources devoted to these existing practices is unlikely to yield the information Washington needs to predict Beijing’s behavior.

To truly gain a grasp of China, the United States needs to think much more creatively about the way it approaches gathering intelligence and the tools it uses to do so. In particular, it must give far more weight to open-source intelligence—which is essential to interpreting Beijing’s thinking—by establishing a centralized office for open-source analysis. It must embrace the most advanced new digital tools to harness and examine the data such intelligence provides. And it should significantly ramp up its efforts to cultivate China expertise and bring more China experts into its ranks. These steps may not give Washington perfect insight into Beijing, but they will make it possible for the United States to sweep up far more information on China and analyze it in a timely manner. They will also improve Washington’s ability to determine how accurate its findings are. And at the very least, when it comes to assessing China’s next moves, they may help prevent Washington from falling into the dark.

HARD TO HANDLE

There is no doubt that China is a vast, complicated, and hard intelligence target. Any spy agency would struggle to understand a strict dictatorship that governs nearly 1.4 billion people. But when it comes to maintaining secrets, China’s size and system can cut both ways. To control its enormous population, China has established 31 provincial-level governments, 299 prefecture-level cities, and over 1,300 counties—each with its own bureaucrats. The CCP also has branches in every major Chinese university, company, and scientific lab. Overall, the CCP itself has 97 million members. This sprawling structure means that, no matter how centralized decision-making is, most of China’s policy objectives and guidance must be communicated openly, creating a call-and-response dynamic that well-positioned analysts can freely observe.

Consider, for example, the path of China’s most recent five-year plan, announced in October 2020. For analysts looking to understand its most important implications, the National People’s Congress revealed very little, publishing only a lengthy but broad outline of what the plan called for. Yet in response to that outline, China’s provinces developed their own, much more detailed five-year plans, each of which contained material that offers substantial insight on the direction of Chinese policy and who is responsible for which part. In the technological domain, these plans helped U.S. intelligence learn where it should try to identify government-backed investment funds, new university partnerships, talent recruitment programs, and the other tools the Chinese government uses to promote innovation.

Washington should study local officials for another reason: they pose a national security threat. It was, for instance, Guangdong Province’s security department that recruited former CIA case officer Jerry Chun Shing Lee in 2010 and then handled him as a source until his arrest in 2018. According to the New York Times, the information Lee provided was one of the reasons China was able to dismantle the CIA’s spy network a decade ago. Guangdong’s security department and other local CCP officials later cultivated a relationship with Australian-Chinese billionaire Chau Chak-wing, who donated more than $2 million to Australian political parties to encourage a pro-CCP stance. According to FBI information publicized in the Australian Parliament, Chau also allegedly facilitated a UN bribery scheme, likely to bring then UN General Assembly President John Ashe into the CCP’s web as part of Beijing’s efforts to change global governance. The bribes resulted in at least one American going to prison for helping make the payments.

The Chinese central government, of course, also tries to cultivate overseas assets. But given the scale of Beijing’s ambitions, even these efforts often happen in plain view. China’s Thousand Talents Program, which works to recruit both expat Chinese and U.S. scientists (in part to gain access to U.S. industrial secrets), has been widely publicized. The country runs hundreds of other publicly known talent recruitment programs, and it boasts roughly 600 overseas recruitment stations. The CCP has multiple international united front groups—some 600 of which operate inside the United States—that support its efforts to obtain expertise and technology from outside countries. Washington is aware of each, but few individuals probably understand the full sweep of these groups’ and programs’ activities.

Any spy agency would struggle to understand a strict dictatorship that governs nearly 1.4 billion people.

The United States has repeatedly said that it is concerned about China stealing its technology. But if Washington wants to better grasp the aims of these Chinese initiatives, it needs to invest more in collecting, processing, and analyzing the large and ever-growing body of public and commercially available information. Procurement and hiring notices, award announcements, and research funding—among many other sources—can all provide useful insights, especially when aggregated. Local CCP apparatchiks also release plenty of easily accessible data through their reports and statements that, studied broadly, can help the United States understand the full breadth of China’s plans and whether they are being carried out.

But so far, Washington’s efforts to use more public information have come up short. Over the last five years, multiple intelligence agencies have established open-source offices, but they are underfunded and do not communicate much with one another. As a result, the insights they gather tend to be siloed and incomplete. To remedy this lack of coordination and improve analysis overall, the United States should create a standalone open-source agency with the authority to acquire, examine, and share open-source data across all parts of the intelligence community.

This entity could be its own agency, or it could be built into one of the existing ones. But it must have an independent voice, one that allows it to influence budgetary decisions and analytic decisions, and generally ensure that open-source perspectives are included in how Washington approaches intelligence. The office should also serve as a gateway between the government and the many open-source analysts who work in the private sector. Ideally, it would employ a workforce of both cleared and uncleared personnel, allowing it to hire experts faster than traditional spy agencies yet still work closely with the intelligence community.

Since open-source findings come from publicly available data, much of the work of a central open-source entity would not be subject to the same restrictions as traditional intelligence agencies and could be shared with allies, preemptively alerting them to Chinese provocations. An open-source agency would also help Washington figure out where it really needs to target its clandestine operations—and where traditional spying is unnecessary. For example, the CCP’s influence efforts have public-facing organizations and meetings that can be tracked and used to identify the officials and groups tied to the CCP, as well as their targets. Clandestine collectors can exploit that knowledge to figure out what actors they should focus on.

FEWER HAYSTACKS, MORE NEEDLES

Creating a dedicated and well-resourced open-source agency is critical to improving U.S. intelligence on Beijing. But it is not sufficient. As the United States gathers more open-source information, it will have more data than any group of analysts can process. In 2017, the director of the National Geospatial Intelligence Agency told an audience that if the United States attempted to manually sift through all the commercial satellite data it will obtain over the next two decades, it would need eight million image analysts. “Even now,” he continued, “every day in just one combat theater with a single sensor, we collect the data equivalent of three NFL seasons, every game. In high definition!”

To process this glut, both the U.S. National Security Commission on Artificial Intelligence and the country’s Special Competitive Studies Project have argued that the intelligence community must embrace artificial intelligence–enabled tools that can identify patterns across huge quantities of data. Several intelligence agencies, including the Office of the Director of National Intelligence, have come up with strategies and plans to do just that. But here, again, Washington’s efforts are insufficient. The agencies’ plans have been unevenly implemented, and various agencies have developed technology that cannot easily exchange information. There are also different data standards across different initiatives, and agencies have struggled to access appropriate computing power. Many employees, afraid of being replaced or making mistakes, have been reluctant to acquire or use new tools.

The Biden administration has pushed the intelligence community to move past this hesitancy. But to ensure that they actually do, the community’s stakeholders at the White House and in Congress will need to see to it that intelligence agencies actually change. Without consistent demand from policymakers, the intelligence community will continue to instead focus on the missions of today, rather than orienting itself for how the U.S.-Chinese rivalry is evolving. The agencies’ leaders must make sure that incoming and existing intelligence officials are trained to use new technologies. It could even make familiarity with any AI tools a prerequisite for being promoted to senior positions.

To drive these changes, the intelligence community should create a unit to run projects that address the bottlenecks that make it hard for agencies to widely adopt digital technologies and artificial intelligence (AI). Its goal should be to build a shared digital architecture for the whole intelligence community, promoting collaboration and making sure that spy agencies can deliver the right information, at the right time, to the right decision-makers. The unit should also make sure this architecture can help deliver findings to foreign partners when needed. In fact, the United States should collaborate with allies as it develops various digital tools. The U.S.-Chinese rivalry crosses the globe, and so Washington must work collaboratively to win.

THE HUMAN FACTOR

Even with rapid access to the best open-source information and advanced technologies, however, the intelligence community will not be able to make reliable assessments of Chinese intentions without the input of the best strategic minds and close students of China. And at present, Washington does not have enough of them. The number of Americans studying China or Chinese has been declining since 2013, and the number of Americans living in that country is also going down. Beijing has become increasingly hostile to foreigners, and so there are fewer job opportunities for Americans in China than there were at the end of the last decade—or even chances for Americans to visit.

Addressing this “knowledge crisis,” as navy intelligence chief Mike Studeman put it in February, will prove challenging. In addition to a dearth of business opportunities, there is no longer any civic institution in China—like the erstwhile University Services Centre in Hong Kong—where U.S. graduate students, professors, government officials, and journalists can go to mingle with their Chinese counterparts. But the intelligence community can make up for this loss by hiring people who lived and worked in China in the past. U.S. agencies will be hesitant to do so; on several occasions, the CIA has discovered that applicants who lived in China were recruited as operatives by Beijing. But if a handful of potential moles have paralyzed the intelligence community’s ability to hire people with experience and knowledge critical to its efforts, then its leaders must modernize the vetting processes.

Intelligence agencies should also consider creating an initiative akin to the military’s Foreign Area Officer program—which trains military officers as country specialists with language proficiency and knowledge about the state’s politics, culture, and society—so that they can cultivate more internal expertise. Such a program would allow employees to join intelligence organizations through normal processes and then later apply to a development program that will turn them into China specialists, including by spending time in-country attached to official diplomatic missions in China (or in Taiwan) and studying Mandarin.

Adding to the intelligence community’s China expertise will improve the U.S. government’s ability to understand the CCP’s intentions, better collect intelligence, and take effective policy action. For example, the United States has underestimated Beijing’s ambitions for years; it was not until 2019 that the Pentagon started saying that the CCP had global ambitions rather than just expanding interests and regional goals. And it took analysts such as the National Intelligence University’s Daniel Tobin, who lived in China and deeply studies the CCP’s documents, to show the government otherwise and to illustrate that the party has had a long and consistent desire to be internationally dominant.

Having this kind of expert enables the United States and allied officials to better anticipate Beijing’s moves. It helps Washington direct the intelligence community’s collectors toward the right operations, such as unraveling the CCP’s influence networks that enable Chinese power internationally, rather than more technical ones—like understanding the specific intent behind the surveillance balloons. More expertise, enabled by better tools and information, will also let U.S. intelligence better support specific actions. The Uyghur Forced Labor Prevention Act, for example, broadly prohibits companies from importing goods forcibly made by members of China’s oppressed Uyghur minority. But enforcement depends on knowing which Chinese companies are participating in the government’s labor programs, which in turn requires that analysts who are proficient enough in Chinese that they can follow evolving terminology, corporate databases, and the ways in which companies and local governments try to hide the telltale signs of involvement.

UP TO THE TASK

The U.S. intelligence community’s struggles with China are certainly serious. But they are not unprecedented. During the Cold War, the United States also faced a large rival governed by a highly secretive communist party. Of course, there are many differences between the Soviet state and today’s China. But then, as now, Washington’s main challenger had a broad global footprint that required U.S. operatives to gather intel from around the world. And then, as now, the two sides worked hard to hunt down moles and plug intelligence vulnerabilities their adversaries exploited.

But as in the twentieth century, the United States can find fresh ways to understand and predict its rival’s behavior. In the Cold War, Washington used new technology—most notably, satellites—to gain information on the Soviet Union. Today, it can use AI to process a growing influx of open-source data. During the twentieth century, the United States was able to rotate analysts through official missions behind the Iron Curtain, or on the Soviet periphery, to gain expertise. It can do the same again in Beijing.

If it takes all these steps, Washington will be able to get a much better handle on China. Indeed, the benefits of some of these innovations have already been shown. Open-source researchers have provided insights into sensitive Chinese activities, such as CCP espionage and political interference abroad. They have helped the United States understand how China organizes and has reformed its electronic signal intelligence, and they have allowed Washington to glean information about Chinese military activity near the Taiwan Strait. If the intelligence community can obtain more open-source intel and embrace AI-enabled tools to examine the data, its analysts would be able to learn—and share—even more. If the community can recruit China experts, it will better anticipate Beijing’s actions and focus analysts’ activities and resources. Beijing’s decision-making may remain opaque, but Washington will still be able to understand China’s behavior.