Social Media Surveillance

Governments are increasingly purchasing sophisticated technology to monitor their citizens’ behavior on social media. Once the preserve of the world’s foremost intelligence agencies, this form of mass surveillance has made its way to a range of countries, from major authoritarian powers to smaller or poorer states that nevertheless hope to track dissidents and persecuted minorities. The booming commercial market for social media surveillance has lowered the cost of entry not only for the security services of dictatorships, but also for national and local law enforcement agencies in democracies, where it is being used with little oversight or accountability. Coupled with an alarming rise in the number of countries where social media users have been arrested for their legitimate online activities, the growing employment of social media surveillance threatens to squeeze the space for civic activism on digital platforms.

A shift to machine-driven monitoring of the public

Social media surveillance refers to the collection and processing of personal data pulled from digital communication platforms, often through automated technology that allows for real-time aggregation, organization, and analysis of large amounts of metadata and content. Broader in scope than spyware, which intercepts communications by targeting specific individuals’ devices, social media surveillance cannot be dismissed as less invasive. Billions of people around the world use these digital platforms to communicate with loved ones, connect with friends and associates, and express their political, social, and religious beliefs. Even when it concerns individuals who seldom interact with such services, the information that is collected, generated, and inferred about them holds tremendous value not only for advertisers, but increasingly for law enforcement and intelligence agencies as well.

Governments have long employed people to monitor speech on social media, including by creating fraudulent accounts to connect with real-life users and gain access to networks. Authorities in Iran have boasted of a 42,000-strong army of volunteers who monitor online speech. Any citizen can report for duty on the Cyber Police (FATA) website. Similarly, the ruling Communist Party in China has recruited thousands of individuals to sift through the internet and report problematic content and accounts to authorities.

Advances in artificial intelligence (AI) have opened up new possibilities for automated mass surveillance. Sophisticated monitoring systems can quickly map users’ relationships through link analysis; assign a meaning or attitude to their social media posts using natural-language processing and sentiment analysis; and infer their past, present, or future locations. Machine learning enables these systems to find patterns that may be invisible to humans, while deep neural networks can identify and suggest whole new categories of patterns for further investigation. Whether accurate or inaccurate, the conclusions made about an individual can have serious repercussions, particularly in countries where one’s political views, social interactions, sexual orientation, or religious faith can lead to closer scrutiny and outright punishment.

The global market for surveillance

The market for social media surveillance has grown, giving intelligence and law enforcement agencies new tools for combing through massive amounts of information. At least 40 of the 65 countries covered by this report have instituted advanced social media monitoring programs.  Moreover, their use by governments is accelerating: in 15 of these countries, it was only in the past year that such programs were either expanded or newly established. Justifying their efforts in the name of enhancing security, limiting disinformation, and ensuring public order, governments have effectively co-opted social media platforms. While these platforms typically present themselves as social connectors and community builders, state agencies in repressive countries see them as vast storehouses of speech and personal information that can be observed, collected, and analyzed to detect and suppress dissent.

China is a leader in developing, employing, and exporting social media surveillance tools. The Chinese firm Semptian has touted its Aegis surveillance system as providing “a full view to the virtual world” with the capacity to “store and analyze unlimited data.” The company claims to be monitoring over 200 million individuals in China—a quarter of the country’s internet users. The company even markets a “national firewall” product, mimicking the so-called Great Firewall that controls internet traffic in China.

Chinese agencies work closely with leading companies to monitor individuals online. A security researcher discovered an unsecured database consisting of the social media profiles, messages, and shared files of some 364 million Chinese users, updated daily, for manual tracking by law enforcement. A complex web of regulations gives the Chinese state access to user content and metadata, allowing authorities to more easily identify and reprimand users who share sensitive content. In March 2019, for example, it was reported that a member of Xinjiang’s persecuted Uighur Muslim minority population was detained and interrogated for three days because someone on his WeChat contact list had “checked in” from Mecca, Saudi Arabia.

Further, several provincial governments in China are reportedly developing a “Police Cloud” system to aggregate data from users’ social media accounts, telecoms records, and e-commerce activity, as well as biometric data and video surveillance footage. The big data policing system can target individuals for interacting with “persons of concern” or for belonging to “certain ethnicities,” a euphemism applying to the Uighur Muslim minority. There, authorities have developed a host of invasive tools, both low- and high-tech, for repressing any behavior that strays from what is acceptable under Xi Jinping Thought—the doctrine of China’s authoritarian leader.

Of the 15 countries in Asia assessed by this report, 13 have social media surveillance programs under development or in use. In Vietnam, the Communist Party government in October 2018 announced a new national surveillance unit equipped with technology to analyze, evaluate, and categorize millions of social media posts. The government has long punished nonviolent activists for what they write on social media; weeks before the October announcement, human rights defender and environmentalist Lê Đình Lượng was convicted and sentenced to 20 years in prison after a one-day trial for trying to overthrow the state, in part for Facebook posts criticizing the government. The new technology will likely enable the government to intensify its crackdown. Meanwhile, Pakistan in February 2019 announced a new social media monitoring program meant to combat extremism, hate speech, and antinational content. Only a month later, the Interior Ministry launched an investigation into journalists and activists who had expressed support for murdered Saudi journalist Jamal Khashoggi on their social media accounts. 

Some countries in Asia are developing their social media surveillance capabilities in close cooperation with US authorities. In September 2018, Philippine officials traveled to North Carolina for training by US Army personnel on developing a new social media monitoring unit. While authorities claim the unit is intended to combat disinformation by violent extremist organizations, the Philippine government’s broad labeling of critical journalists and users as terrorists suggests that monitoring efforts will extend far beyond any legitimate security threat. Bangladesh’s Rapid Action Battalion (RAB) was approved to travel to the United States in April 2019 to receive training on “Location Based Social Network Monitoring System Software.” The RAB, which is infamous for human rights violations including extrajudicial killings, enforced disappearances, and torture, was given 1.2 billion taka ($14 million) by the Bangladeshi government for “state-of-the-art equipment” to monitor in real time what it considers to be rumors and propaganda. These developments occurred in a year when authorities led a violent crackdown on dissent during national protests and general elections.

The Middle East and North Africa region, home to some of the world’s most repressive regimes, is also a booming market for social media surveillance. Companies scheduled to attend a Dubai trade show in 2020 represent countries including China, India, Israel, Italy, the United States, and the United Kingdom. Knowlesys, a Chinese company whose clients reportedly include the Chinese military and government bodies, will hold live demonstrations on how to “monitor your targets’ messages, profiles, locations, behaviors, relationships, and more,” and how to “monitor public opinion for election.” Semptian, which has clients in the region, has a price range of $1.5 million to $2.5 million for monitoring the online activities of a population of five million people—an affordable price for most dictators.

In December 2018, it was reported that Kazakhstan had purchased a $4.3 million automated monitoring tool to track signs of political discontent on social media. The firm supplying the software is linked to Russia’s Federal Security Service and has been subjected to sanctions by the US Treasury Department for its activities surrounding the 2016 US elections. Screenshots revealed that the product uses deep learning to “detect materials that discredit the state.” The tools could easily be abused in Kazakhstan, where individuals have received multiyear prison sentences for social media posts that are deemed supportive of the Democratic Choice of Kazakhstan, a banned opposition party. 

Russia has used sophisticated social media surveillance tools for many years. The government issued three tenders in 2012 for the development of research methods related to “social networks intelligence,” “tacit control on the internet,” and “a special software package for the automated dissemination of information in large social networks,” foreshadowing how intelligence agencies would eventually master the manipulation of social media at home and abroad. This May, authorities released a tender for technology to collect, analyze, and conduct sentiment analysis on social media content relating to President Vladimir Putin and other topics of interest to the government. The year featured more protest-related arrests, internet shutdowns, and legal restrictions in Russia, suggesting that any new monitoring technology would simply add to the government’s arsenal of tools for clamping down on unauthorized political mobilization.

Monitoring projects are under way in Africa as well. The government of Nigeria allocated 2.2 billion naira ($6.6 million) in its 2018 budget for a “Social Media Mining Suite,” having already ordered the military to watch for antigovernment content online. In an ominous sign, the country experienced an increase in arrests for internet activity over the past year. Human rights and democracy activist Ibrahim Garba Wala, known as IG Wala, was sentenced in April to 12 years in prison for criminal defamation, public incitement, and unlawful assembly; the charges stemmed from Facebook posts alleging corruption in the National Hajj Commission. Israeli firms Verint and WebIntPro have reportedly sold similar surveillance software to Angola and Kenya, respectively.

In strong democracies, new tools of potential repression

The social media surveillance tools that have appeared in democracies got their start on foreign battlefields and in counterterrorism settings, designed to monitor acute security threats in places like Syria. Many US data-mining companies received seed money from the Central Intelligence Agency through its In-Q-Tel venture capital fund. While authorities in the past typically justified the use of these tools with the need to combat serious crimes such as terrorism, child sexual abuse, and large-scale narcotics trafficking, law enforcement and other agencies at the local, state, and federal levels are increasingly repurposing them for more questionable practices, such as screening travelers for their political views, tracking students’ behavior, or monitoring activists and protesters. This expansion makes oversight of surveillance policies more difficult and raises the risk that constitutionally protected activities will be impaired.

For example, in the United States, agencies within the Department of Homeland Security (DHS)—including Customs and Border Protection (CBP), Citizenship and Immigration Services, and Immigration and Customs Enforcement (ICE)—have used automated technology to collect and analyze personal information, with limited oversight and transparency. By claiming that its power to conduct warrantless searches extends within a 100-mile radius of any US border, DHS has effectively asserted extrajudicial surveillance powers over 200 million people. CBP has even purchased technology from Cellebrite, an Israeli company, to bypass encryption and passwords and enable quick extraction of data from phones and computers, including social media content. There has been a spike in device searches at the borders in recent years; the number of such searches, normally limited under the Fourth Amendment of the constitution, increased by 292 percent, from 8,503 to 33,295, between fiscal year 2015 and fiscal year 2018. Over that same period, inbound travel to the United States increased by less than 3 percent.

These searches have become part of the government’s drive toward big data surveillance. The resulting information is frequently deposited in massive multiagency databases where it can be combined with public records, secret intelligence materials, and datasets (including social media data) assembled by private companies. In one case, ICE paid the data analytics company Palantir $42.3 million for a one-year contract related to FALCON, a custom-built database management tool. Its “Search and Analysis System” enables agents to analyze trends and establish links between individuals based on information gathered during border searches, purchased from private data brokers, and obtained from other intelligence collection exercises. Similar tools developed by Palantir are used by some 300 police departments in the state of California alone, as well as by police forces in Chicago, Los Angeles, New Orleans, and New York City. Many of these programs are facilitated through DHS and its Regional Intelligence Centers.

The consequences of government intrusion into the digital public square

For authoritarian and democratic governments alike, the potential for abuse presented by advanced social media surveillance is staggering. In 2019, Freedom House found that 47 of the 65 countries assessed featured arrests of users for political, social, or religious speech—a record high. The blanket monitoring of online activities for undesirable or illegal speech will undoubtedly lead to more arrests, particularly in environments that lack strong protections for free expression. Monitoring designed to detect and deter protests will also help stifle democracy movements in authoritarian settings.

Even in countries with considerable safeguards for fundamental freedoms, there are already reports of abuse. In the United Kingdom, for example, London police reportedly monitored nearly 9,000 activists from across the political spectrum—many of whom had no criminal background—using geolocation tracking and sentiment analysis on data scraped from Facebook, Twitter, and other platforms. This information was then compiled in secret dossiers on each campaigner. Similar dynamics are evident in the United States, where leaked documents revealed in March 2019 that CBP had created a list of 59 US and foreign immigration activists, journalists, lawyers, and Facebook group administrators who should be targeted for greater scrutiny at the US-Mexico border, leading to arrests in nine cases. ICE has also monitored social media in New York City to gather information on groups protesting the administration’s immigration and gun-control policies. Such profiling poses a distinct threat to basic civil liberties. As the US Supreme Court ruled in 1958, “inviolability of privacy in group association may in many circumstances be indispensable to preservation of freedom of association, particularly where a group espouses dissident beliefs.”

The chilling effect on free expression caused by increased surveillance is well documented. Activists and journalists who might otherwise hold governments to account for wrongdoing are more inclined to self-censor, while dissidents and members of marginalized communities will think twice about discussing their political opinions online to avoid arrests or travel restrictions. Furthermore, social media monitoring designed to quell mobilization and identify protesters hinders the public’s ability to use online tools to associate and assemble peacefully. Finally, indiscriminate monitoring of the general population’s online communications—even when those communications are nominally public—runs afoul of due process standards enshrined in democratic constitutions and international human rights law.

Protecting human rights in the age of AI surveillance

There is little if any public evidence that such technology is more effective than less-invasive alternatives for ensuring national security and combating serious crimes. Social media activity such as original content, likes, or shares—particularly speech that is rendered in slang or languages other than English—is susceptible to misinterpretation and misclassification. Research has estimated the accuracy rates of natural-language processing tools at 70 to 80 percent. While they are often justified as a means to reduce human error, algorithmic tools can further entrench racial or religious discrimination due to reliance on inaccurate or biased data. The resulting false positives can add innocent people to government watch lists, often without their knowledge, leaving them with little recourse for remedying the mistake.

At the very least, social media surveillance must come under greater oversight. The use of such programs must be transparent, including sustained dialogue between law enforcement and affected communities. Public civil rights assessments should be conducted, and authorities should be held accountable when tools are misused and offer remedies for any victims. Online surveillance technology should not be used to proactively monitor the planning and organization of peaceful protest activities or individuals’ involvement in nonviolent political groups. And governments should swiftly amend existing privacy legislation to address the proper use of this technology.

Thanks to the development of AI-assisted tools, governments now have a greater capacity for surveillance than ever before. Given their potential impact on fundamental rights, policymakers and citizens must ask themselves whether these new tools are necessary or desirable in a democratic society. It is time to move beyond outdated arguments that individuals “should have nothing to hide” or do not have a reasonable expectation of privacy in public areas. The survival of democracy requires vibrant public spaces, both offline and online, where individuals can collaborate, organize, and go about their personal lives without fear of constant surveillance.

Hero Image Caption: A police officer monitors various social media channels at the Punjab Police Integrated Command, Control and Communication Center (IC3) in Lahore, Pakistan. Photo Credit: Asad Zaidi/Bloomberg via Getty Images.