12 February 2020

The Digital Dictators

By Andrea Kendall-Taylor, Erica Frantz, and Joseph Wright

The Stasi, East Germany’s state security service, may have been one of the most pervasive secret police agencies that ever existed. It was infamous for its capacity to monitor individuals and control information flows. By 1989, it had almost 100,000 regular employees and, according to some accounts, between 500,000 and two million informants in a country with a population of about 16 million. Its sheer manpower and resources allowed it to permeate society and keep tabs on virtually every aspect of the lives of East German citizens. Thousands of agents worked to tap telephones, infiltrate underground political movements, and report on personal and familial relationships. Officers were even positioned at post offices to open letters and packages entering from or heading to noncommunist countries. For decades, the Stasi was a model for how a highly capable authoritarian regime could use repression to maintain control.

In the wake of the apparent triumph of liberal democracy after the Cold War, police states of this kind no longer seemed viable. Global norms about what constituted a legitimate regime had shifted. At the turn of the millennium, new technologies, including the Internet and the cell phone, promised to empower citizens, allowing individuals greater access to information and the possibility to make new connections and build new communities.


But this wishful vision of a more democratic future proved naive. Instead, new technologies now afford rulers fresh methods for preserving power that in many ways rival, if not improve on, the Stasi’s tactics. Surveillance powered by artificial intelligence (AI), for example, allows despots to automate the monitoring and tracking of their opposition in ways that are far less intrusive than traditional surveillance. Not only do these digital tools enable authoritarian regimes to cast a wider net than with human-dependent methods; they can do so using far fewer resources: no one has to pay a software program to monitor people’s text messages, read their social media posts, or track their movements. And once citizens learn to assume that all those things are happening, they alter their behavior without the regime having to resort to physical repression.

This alarming picture stands in stark contrast to the optimism that originally accompanied the spread of the Internet, social media, and other new technologies that have emerged since 2000. Such hopefulness peaked in the early 2010s as social media facilitated the ouster of four of the world’s longest-ruling dictators, in Egypt, Libya, Tunisia, and Yemen. In a world of unfettered access to information and of individuals empowered by technology, the argument went, autocrats would no longer be able to maintain the concentration of power that their systems depend on. It’s now clear, however, that technology does not necessarily favor those seeking to make their voices heard or stand up to repressive regimes. Faced with growing pressure and mounting fear of their own people, authoritarian regimes are evolving. They are embracing technology to refashion authoritarianism for the modern age.

Led by China, today’s digital autocracies are using technology—the Internet, social media, AI—to supercharge long-standing authoritarian survival tactics. They are harnessing a new arsenal of digital tools to counteract what has become the most significant threat to the typical authoritarian regime today: the physical, human force of mass antigovernment protests. As a result, digital autocracies have grown far more durable than their pre-tech predecessors and their less technologically savvy peers. In contrast to what technology optimists envisioned at the dawn of the millennium, autocracies are benefiting from the Internet and other new technologies, not falling victim to them.
THE SPECTER OF PROTEST

The digital age changed the context in which authoritarian regimes operate. Such new technologies as the Internet and social media reduced barriers to coordination, making it easier for ordinary citizens to mobilize and challenge unresponsive and repressive governments. Data from the Mass Mobilization Project, compiled the political scientists David Clark and Patrick Regan, and the Autocratic Regimes data set, which two of us (Erica Frantz and Joseph Wright) have helped build, reveal that between 2000 and 2017, 60 percent of all dictatorships faced at least one antigovernment protest of 50 participants or more. Although many of these demonstrations were small and posed little threat to the regime, their sheer frequency underscores the continuous unrest that many authoritarian governments face.

Many of these movements are succeeding in bringing about the downfall of authoritarian regimes. Between 2000 and 2017, protests unseated ten autocracies, or 23 percent of the 44 authoritarian regimes that fell during the period. Another 19 authoritarian regimes lost power via elections. And while there were nearly twice as many regimes ousted by elections as by protests, many of the elections had followed mass protest campaigns.

The rise in protests marks a significant change in authoritarian politics. Historically, coups by military elites and officers posed the greatest threat to dictatorships. Between 1946 and 2000, coups ousted roughly a third of the 198 authoritarian regimes that collapsed in that period. Protests, in contrast, unseated far fewer, accounting for about 16 percent of that total. Fast-forward to this century, and a different reality emerges: coups unseated around nine percent of the dictatorships that fell between 2001 and 2017, while mass movements led to the toppling of twice as many governments. In addition to toppling regimes in the Arab Spring, protests led to the ouster of dictatorships in Burkina Faso, Georgia, and Kyrgyzstan. Protests have become the most significant challenge that twenty-first-century authoritarian regimes face.


Police on their way to contain anti-government protests in Phnom Penh, Cambodia, September 2013Justin Mott / The New York Times ​/ Redux

The growing threat of protests has not been lost on today’s autocrats. In the past, when they feared coups, most such leaders relied on “coup proofing” tactics, such as overpaying the security services to win their loyalty or rotating elites through positions of power so that no one could develop an independent base of support. As protests have increased, however, authoritarian regimes have adapted their survival tactics to focus on mitigating the threat from mass mobilization. Data compiled by Freedom House reveal that since 2000, the number of restrictions on political and civil liberties globally has grown. A large share of this increase has occurred in authoritarian countries, where leaders impose restrictions on political and civil liberties to make it harder for citizens to organize and agitate against the state.

Beyond narrowing the space for civil society, authoritarian states are also learning to use digital tools to quell dissent. Although technology has helped facilitate protests, today’s digitally savvy authoritarian regimes are using some of the same technological innovations to push back against dangerous popular mobilizations.
MEANS OF CONTROL

Our analysis using data from Varieties of Democracy’s data set (which covers 202 countries) and the Mass Mobilization Project shows that autocracies that use digital repression face a lower risk of protests than do those autocratic regimes that do not employ these same tools. Digital repression not only decreases the likelihood that a protest will occur but also reduces the chances that a government will face large, sustained mobilization efforts, such as the “red shirt” protests in Thailand in 2010 or the anti-Mubarak and antimilitary protests in Egypt in 2011. The example of Cambodia illustrates how these dynamics can play out.

The government of Prime Minister Hun Sen, who has been in office since 1985, has adopted technological methods of control to help maintain its grip on power. Under Hun Sen’s rule, traditional media have restricted their coverage of the Cambodian opposition. In the run-up to the July 2013 election, this led the opposition to rely heavily on digital tools to mobilize its supporters. The election was fraudulent, prompting thousands of citizens to take to the streets to demand a new vote. In addition to employing brute force to quell the protests, the government ratcheted up its use of digital repression. For instance, in August 2013, one Internet service provider temporarily blocked Facebook, and in December 2013, authorities in the province of Siem Reap closed down more than 40 Internet cafés. The following year, the government announced the creation of the Cyber War Team, tasked with monitoring the Internet to flag antigovernment activity online. A year later, the government passed a law giving it broad control over the telecommunications industry and established an enforcement body that could suspend telecommunications firms’ services and even fire their staff. Partly as a result of these steps, the protest movement in Cambodia fizzled out. According to the Mass Mobilization Project, there was only one antigovernment protest in the country in 2017, compared with 36 in 2014, when the opposition movement was at its peak.

Dictatorships harness technology not only to suppress protests but also to stiffen older methods of control. Our analysis drawing from Varieties of Democracy’s data set suggests that dictatorships that increase their use of digital repression also tend to increase their use of violent forms of repression “in real life,” particularly torture and the killing of opponents. This indicates that authoritarian leaders don’t replace traditional repression with digital repression. Instead, by making it easier for authoritarian regimes to identify their opposition, digital repression allows them to more effectively determine who should get a knock on the door or be thrown in a cell. This closer targeting of opponents reduces the need to resort to indiscriminate repression, which can trigger a popular backlash and elite defections.
THE CHINA MODEL

The advancement of AI-powered surveillance is the most significant evolution in digital authoritarianism. High-resolution cameras, facial recognition, spying malware, automated text analysis, and big-data processing have opened up a wide range of new methods of citizen control. These technologies allow governments to monitor citizens and identify dissidents in a timely—and sometimes even preemptive—manner.

No regime has exploited the repressive potential of AI quite as thoroughly as the one in China. The Chinese Communist Party collects an incredible amount of data on individuals and businesses: tax returns, bank statements, purchasing histories, and criminal and medical records. The regime then uses AI to analyze this information and compile “social credit scores,” which it seeks to use to set the parameters of acceptable behavior and improve citizen control. Individuals or companies deemed “untrustworthy” can find themselves excluded from state-sponsored benefits, such as deposit-free apartment rentals, or banned from air and rail travel. Although the CCP is still honing this system, advances in big-data analysis and decision-making technologies will only improve the regime’s capacity for predictive control, what the government calls “social management.”

China demonstrates how digital repression aids physical repression.

China also demonstrates the way digital repression aids the physical variety—on a mass scale. In Xinjiang, the Chinese government has detained more than a million Uighurs in “reeducation” camps. Those not in camps are stuck in cities where neighborhoods are surrounded by gates equipped with facial recognition software. That software determines who may pass, who may not, and who will be detained on sight. China has collected a vast amount of data on its Uighur population, including cell phone information, genetic data, and information about religious practices, which it aggregates in an attempt to stave off actions deemed harmful to public order or national security.

New technologies also afford Chinese officials greater control over members of the government. Authoritarian regimes are always vulnerable to threats from within, including coups and high-level elite defections. With the new digital tools, leaders can keep tabs on government officials, gauging the extent to which they advance regime objectives and rooting out underperforming officials who over time can tarnish public perception of the regime. For example, research has shown that Beijing avoids censoring citizens’ posts about local corruption on Weibo (the Chinese equivalent of Twitter) because those posts give the regime a window into the performance of local officials.

Facial recognition technology in use at a railway station in Wuhan, China, August 2017Xiong Qi / Xinhua / Eyevine / Redu​x

In addition, the Chinese government deploys technology to perfect its systems of censorship. AI, for example, can sift through massive amounts of images and text, filtering and blocking content that is unfavorable to the regime. As a protest movement heated up in Hong Kong last summer, for example, the Chinese regime simply strengthened its “Great Firewall,” removing subversive content from the Internet in mainland China almost instantaneously. And even if censorship fails and dissent escalates, digital autocracies have an added line of defense: they can block all citizens’ access to the Internet (or large parts of it) to prevent members of the opposition from communicating, organizing, or broadcasting their messages. In Iran, for example, the government successfully shut down the Internet across the country amid widespread protests last November.

Although China is the leading player in digital repression, autocracies of all stripes are looking to follow suit. The Russian government, for example, is taking steps to rein in its citizens’ relative freedom online by incorporating elements of China’s Great Firewall, allowing the Kremlin to cut off the country’s Internet from the rest of the world. Likewise, Freedom House reported in 2018 that several countries were seeking to emulate the Chinese model of extensive censorship and automated surveillance, and numerous officials from autocracies across Africa have gone to China to participate in “cyberspace management” training sessions, where they learn Chinese methods of control.
THE VELVET GLOVE

Today’s technologies not only make it easier for governments to repress critics; they also make it easy to co-opt them. Tech-powered integration between government agencies allows the Chinese regime to more precisely control access to government services, so that it can calibrate the distribution—or denial—of everything from bus passes and passports to jobs and access to education. The nascent social credit system in China has the effect of punishing individuals critical of the regime and rewarding loyalty. Citizens with good social credit scores benefit from a range of perks, including expedited overseas travel applications, discounted energy bills, and less frequent audits. In this way, new technologies help authoritarian regimes fine-tune their use of reward and refusal, blurring the line between co-option and coercive control.

Dictatorships can also use new technologies to shape public perception of the regime and its legitimacy. Automated accounts (or “bots”) on social media can amplify influence campaigns and produce a flurry of distracting or misleading posts that crowd out opponents’ messaging. This is an area in which Russia has played a leading role. The Kremlin floods the Internet with pro-regime stories, distracting online users from negative news, and creates confusion and uncertainty through the spread of alternative narratives.

Maturing technologies such as so-called microtargeting and deepfakes—digital forgeries impossible to distinguish from authentic audio, video, or images—are likely to further boost the capacity of authoritarian regimes to manipulate their citizens’ perceptions. Microtargeting will eventually allow autocracies to tailor content for specific individuals or segments of society, just as the commercial world uses demographic and behavioral characteristics to customize advertisements. AI-powered algorithms will allow autocracies to microtarget individuals with information that either reinforces their support for the regime or seeks to counteract specific sources of discontent. Likewise, the production of deepfakes will make it easier to discredit opposition leaders and will make it increasingly difficult for the public to know what is real, sowing doubt, confusion, and apathy.

Digital tools might even help regimes make themselves appear less repressive and more responsive to their citizens. In some cases, authoritarian regimes have deployed new technologies to mimic components of democracy, such as participation and deliberation. Some local Chinese officials, for example, are using the Internet and social media to allow citizens to voice their opinions in online polls or through other digitally based participatory channels. A 2014 study by the political scientist Rory Truex suggested that such online participation enhanced public perception of the CCP among less educated citizens. Consultative sites, such as the regime’s “You Propose My Opinion” portal, make citizens feel that their voices matter without the regime having to actually pursue genuine reform. By emulating elements of democracy, dictatorships can improve their attractiveness to citizens and deflate the bottom-up pressure for change.
DURABLE DIGITAL AUTOCRACIES

As autocracies have learned to co-opt new technologies, they have become a more formidable threat to democracy. In particular, today’s dictatorships have grown more durable. Between 1946 and 2000—the year digital tools began to proliferate—the typical dictatorship ruled for around ten years. Since 2000, this number has more than doubled, to nearly 25 years.

Not only has the rising tide of technology seemingly benefited all dictatorships, but our own empirical analysis shows that those authoritarian regimes that rely more heavily on digital repression are among the most durable. Between 2000 and 2017, 37 of the 91 dictatorships that had lasted more than a year collapsed; those regimes that avoided collapse had significantly higher levels of digital repression, on average, than those that fell. Rather than succumb to what appeared to be a devastating challenge to their power—the emergence and spread of new technologies—many dictatorships leverage those tools in ways that bolster their rule.

Archived files of the East German secret police, or Stasi, in Berlin, Germany, October 2019Stefan Boness /Panos Pictures / R​edux

Although autocracies have long relied on various degrees of repression to support their objectives, the ease with which today’s authoritarian regimes can acquire this repressive capacity marks a significant departure from the police states of the past. Building the effectiveness and pervasiveness of the East German Stasi, for example, was not something that could be achieved overnight. The regime had to cultivate the loyalty of thousands of cadres, training them and preparing them to engage in on-the-ground surveillance. Most dictatorships simply do not have the ability to create such a vast operation. There was, according to some accounts, one East German spy for every 66 citizens. The proportion in most contemporary dictatorships (for which there are data) pales in comparison. It is true that in North Korea, which ranks as possibly the most intense police state in power today, the ratio of internal security personnel and informants to citizens is 1 to 40—but it was 1 to 5,090 in Iraq under Saddam Hussein and 1 to 10,000 in Chad under Hissène Habré. In the digital age, however, dictatorships don’t need to summon immense manpower to effectively surveil and monitor their citizens.

Instead, aspiring dictatorships can purchase new technologies, train a small group of officials in how to use them—often with the support of external actors, such as China—and they are ready to go. For example, Huawei, a Chinese state-backed telecommunications firm, has deployed its digital surveillance technology in over a dozen authoritarian regimes. In 2019, reports surfaced that the Ugandan government was using it to hack the social media accounts and electronic communications of its political opponents. The vendors of such technologies don’t always reside in authoritarian countries. Israeli and Italian firms have also sold digital surveillance software to the Ugandan regime. Israeli companies have sold espionage and intelligence-gathering software to a number of authoritarian regimes across the world, including Angola, Bahrain, Kazakhstan, Mozambique, and Nicaragua. And U.S. firms have exported facial recognition technology to governments in Saudi Arabia and the United Arab Emirates.
A SLIPPERY SLOPE

As autocracies last longer, the number of such regimes in place at any point in time is likely to increase, as some countries backslide on democratic rule. Although the number of autocracies globally has not risen substantially in recent years, and more people than ever before live in countries that hold free and fair elections, the tide may be turning. Data collected by Freedom House show, for example, that between 2013 and 2018, although there were three countries that transitioned from “partly free” to “free” status (the Solomon Islands, Timor-Leste, and Tunisia), there were seven that experienced the reverse, moving from a status of “free” to one of “partly free” (the Dominican Republic, Hungary, Indonesia, Lesotho, Montenegro, Serbia, and Sierra Leone).

The risk that technology will usher in a wave of authoritarianism is all the more concerning because our own empirical research has indicated that beyond buttressing autocracies, digital tools are associated with an increased risk of democratic backsliding in fragile democracies. New technologies are particularly dangerous for weak democracies because many of these digital tools are dual use: technology can enhance government efficiency and provide the capacity to address challenges such as crime and terrorism, but no matter the intentions with which governments initially acquire such technology, they can also use these tools to muzzle and restrict the activities of their opponents.

Pushing back against the spread of digital authoritarianism will require addressing the detrimental effects of new technologies on governance in autocracies and democracies alike. As a first step, the United States should modernize and expand legislation to help ensure that U.S. entities are not enabling human rights abuses. A December 2019 report by the Center for a New American Security (where one of us is a senior fellow) highlights the need for Congress to restrict the export of hardware that incorporates AI-enabled biometric identification technologies, such as facial, voice, and gait recognition; impose further sanctions on businesses and entities that provide surveillance technology, training, or equipment to authoritarian regimes implicated in human rights abuses; and consider legislation to prevent U.S. entities from investing in companies that are building AI tools for repression, such as the Chinese AI company SenseTime.

The United States must lead in AI in ways that are consistent with its democratic values.

The U.S. government should also use the Global Magnitsky Act, which allows the U.S. Treasury Department to sanction foreign individuals involved in human rights abuses, to punish foreigners who engage in or facilitate AI-powered human rights abuses. CCP officials responsible for atrocities in Xinjiang are clear candidates for such sanctions.

U.S. government agencies and civil society groups should also pursue actions to mitigate the potentially negative effects of the spread of surveillance technology, especially in fragile democracies. The focus of such engagement should be on strengthening the political and legal frameworks that govern how surveillance technologies are used and building the capacity of civil society and watchdog organizations to check government abuse.

What is perhaps most critical, the United States must make sure it leads in AI and helps shape global norms for its use in ways that are consistent with democratic values and respect for human rights. This means first and foremost that Americans must get this right at home, creating a model that people worldwide will want to emulate. The United States should also work in conjunction with like-minded democracies to develop a standard for digital surveillance that strikes the right balance between security and respect for privacy and human rights. The United States will also need to work closely with like-minded allies and partners to set and enforce the rules of the road, including by restoring U.S. leadership in multilateral institutions such as the United Nations.

AI and other technological innovations hold great promise for improving everyday lives, but they have indisputably strengthened the grip of authoritarian regimes. The intensifying digital repression in countries such as China offers a bleak vision of ever-expanding state control and ever-shrinking individual liberty.

But that need not be the only vision. In the near term, rapid technological change will likely produce a cat-and-mouse dynamic as citizens and governments race to gain the upper hand. If history is any guide, the creativity and responsiveness of open societies will in the long term allow democracies to more effectively navigate this era of technological transformation. Just as today’s autocracies have evolved to embrace new tools, so, too, must democracies develop new ideas, new approaches, and the leadership to ensure that the promise of technology in the twenty-first century doesn’t become a curse.

No comments: