US global surveillance reinforces racial and social inequalities- companies decrease user privacy for government intelligence
Michael Kwet 17 Yale University, Visiting Fellow Rhodes University, Ph.D. candidate;Operation Phakisa Education: Why a secret? Mass surveillance, inequality, and race in South Africa's emerging national e-education system; https://firstmonday.org/ojs/index.php/fm/article/view/8054;
Today, the United States is at the center of a new, digital global surveillance apparatus. U.S. whistleblowers like William Binney, Thomas Drake, Kirk Wiebe, and Ed Lumis revealed that the U.S. government is conducting bulk surveillance on foreign and domestic populations. Their claims were confirmed by Edward Snowden, who leaked a massive trove of documents to journalists in June 2013. Various leaks and whistleblowers have revealed that the United States government partners with major U.S.-based technology corporations to conduct mass surveillance on domestic and global populations. The National Security Agency (NSA) is at the center of the new surveillance regime (Greenwald, 2014; Granick, 2017). The vast scope of NSA surveillance is indicated by various programs disclosed by whistleblowers. The two primary tools used to conduct surveillance are Upstream (which taps the Internet backbone, often with direct cooperation from U.S. corporations) and PRISM (which gives the NSA access to data from major corporate service providers such as Facebook, Google, Microsoft, and Skype) (Macaskill and Dance, 2013). A program called XKEYSCORE is used to collect “nearly everything a user does on the Internet” and provide them with a Google-like search interface for surveillance targets (Greenwald, 2013; Dorling, 2013). Through BULLRUN, the NSA inserted vulnerabilities into popular protocols like HTTPS, voice-over-IP, and Secure Sockets Layer (SSL) (Ball, et al., 2013) [29]. The NSA permits it to store any and all encrypted communications for as long as they wish, in hopes they may one day decrypt the content (Perloth, et al., 2013). The NSA’s new $2 billion facility in Bluffdale, Utah has four 25,000-square-foot data halls to store “rows and rows” of servers (Bamford, 2012). According to NSA whistleblower William Binney, Bluffdale was built to expand the NSA’s data storage capacity. He estimates “just for e-mails and phone calls alone they have approaching 20 trillion transactions of US citizens with other US citizens” (Binney, 2013). Banking, social networking data, and other services add trillions of more transactions. The mass interception and collection of electronic communications demonstrates that the United States and its allies aim to bring nearly all electronic communications under its vast global surveillance net. There are recent examples of Western intelligence agencies spying on South Africans. In June 2014, the Washington Post reported that the NSA provided legal certification “to intercept through U.S. companies not just the communications of its overseas targets but any communications about its targets as well” (Nakashima and Gellman, 2014). One hundred ninety three countries are certified targets, including South Africa (Washington Post, 2014). In 2014, Der Spiegel published slides detailing an NSA program, TREASUREMAP, which aims to provide a “near real-time, interactive map of the global internet” that plots every device connected in “a kind of Google Earth for global data traffic, a bird’s eye view of the planet’s digital arteries” (Müller-Maguhn, et al., 2014). The NSA includes South Africa among its “13 covered servers in unwitting data centers around the globe” (National Security Agency [NSA], n.d.). The data centers help the NSA “perform [DNS] monitoring covertly and to get a more global view on DNS name resolution and service availability” [30]. In 2016, Le Monde reported that Britain’s Government Communications Headquarters (GCHQ), an NSA ally, tapped into telecommunications networks by targeting their managers. South Africa-based telecommunications corporation MTN Group was among those under surveillance (Leloup and Utersinger, 2016). The GCHQ also aimed to retrieve the briefings of the South African delegates to G20 and G8 summit meetings (Guardian, 2013). In 2015, the United Kingdom Investigatory Powers Tribunal found the GCHQ unlawfully breached the European Convention on Human Rights for spying on the South African-based Legal Resources Centre (LRC) (RDM Newswire, 2015). The LRC is known for its work protecting human rights. Within South Africa there is suspicion that domestic corporations and the state are conducting bulk surveillance. In 2011, Wikileaks published leaked documents from Stellenbosch-based VASTech SA Pty Ltd. showcasing equipment for bulk communications interception. The content of phone calls and e-mail messages, as well as metadata from telephone and Internet communications, can be swept up by VASTech’s Zebra, Badger, and Satellite Signal Analyser products (Swart, 2016a). The equipment was sold to Muammar Gaddafi, who used it for bulk surveillance of the Libyan population (Sonne and Coker, 2011). In October 2016, The Intercept published a leaked document revealing that VASTech has since developed more powerful “NSA-like” equipment it claims “rivals a government spy agency” (McLaughlin, 2016). During the mid-2000s, VASTech was subsidized by the South African government (Mchunu, 2013). It is not known if the government owns or uses VASTech’s mass surveillance equipment. However, the evidence suggests it desires to conduct bulk surveillance to counter political dissent. In June 2016, investigative journalist Heidi Swart reported contents of a leaked document, “The 2014 National Intelligence Priorities”. According to Swart, “The document states that countering the ‘most serious threats to our national security’ that require ‘immediate and sustained intelligence collection” involves the ‘maximum use of covert human and technical means’.’ The “intelligence gathering is justified by several ‘threats to national security’. For example, one intelligence priority for 2014 centered on national and provincial elections. The SSA Domestic Branch was tasked with investigating and counterplanning for a ‘so-called “Arab Spring” uprising prior to elections’”. “Another area of interest for the SAA,” Swart added, “was investigating and counterplanning for ‘violent industrial action’, with the investigation of ‘inter-union rivalry as a driver of violent industrial action’ high on the agenda. The transport industry was to be watched to allow for ‘counterplanning for violence in the taxi industry’”. Areas of surveillance include “terrorism, espionage, violence against foreign nationals, gangsterism, and the private security industry” (Swart, 2016b). Privacy advocates consider South African communications legislation too weak to adequately safeguard privacy (Right2Know, 2014). Under the Regulation of Interception of Communications and Provision Act (RICA), telecommunications service providers must deploy infrastructure to allow the monitoring and interception of communications, while Internet service providers are required to store citizen metadata for three–five years. Legislation fails to regulate the most powerful mass surveillance capabilities conducted at the National Communications Centre, and citizens are required to register SIM cards in a national database (Right2Know, 2016). The “2014 National Priorities” document and other facts about surveillance equipment and legislation make clear that all South African inhabitants face the threat of mass government surveillance. As digital technology spreads in South Africa, its inhabitants must also grapple with ubiquitous surveillance captured for big data analytics via the private sector. There are warning signs that this will have deleterious effects on civil rights and liberties. In May 2016, Angwin, et al. published a case study for ProPublica in which the Correctional Offender Management Profiling for Alternative Sanctions (COMPAS) computer software created by Northpointe corporation for the U.S. criminal justice system assigned a higher criminal “risk assessment” to blacks than to whites. ProPublica’s study assessed 7,000 people arrested in Florida. They found the risk assessment algorithm used to predict violent crime flagged black defendants as future criminals at twice the rate as white defendants. Northpointe’s algorithm is proprietary. As such, there is no way for the public to understand how it gave rise to this outcome. However, some details were provided to ProPublica. Northpointe does not categorize individuals by race, but they do assess factors correlated to race, such as educational degree and employment status. Northpointe’s founder, Tim Brennan, maintains that categories correlated to race like poverty, joblessness, and social marginalization must be included lest “your ... accuracy goes down” (Angwin, et al., 2016). Their predictive policing software is among the most widely used assessment tools in the U.S. criminal justice system. Commercial surveillance also threatens to reinforce socioeconomic inequality. For example, companies are seeking to exploit novel forms of data to determine credit scores. Social networking services like Facebook seek to determine creditworthiness based on “social credit scores” (data from social networking) (Packin and Lev-Aretz, 2016). A person’s Facebook friends, for example, might better indicate her creditworthiness than her financial history derived from a FICO score. As with Northpointe’s algorithm, facially neutral categories might adversely affect oppressed groups. In the case of race, if having personal wealth and wealthy friends raises your credit score, and white people have more wealth, then the algorithm would favor whites and reinforce race-based inequality [31]. Socioeconomic inequality is all the more prone to insidious reinforcement when race-associated attributes tacitly emerge through data analytics. This is sometimes called “latent trait inference” whereby algorithmic processes are able to accurately infer traits from data even if they are not disclosed. Facebook has been dancing around this issue. For example, their users do not disclose their race (there is no “race” category), but Facebook offers advertisers “ethnic affinity” categories such as African American, Asian American, and Hispanic. Given that any user from any “race” can fall into the category based on “indicators” (e.g., a white person who is a member of the African American Chamber of Commerce will help define their “ethnic affinity”), Facebook states they are not racially profiling (Newtiz, 2016). In March 2016, Business Insider revealed different users received different trailers for the film Straight Outta Compton based on their Facebook affinity group (McAlone, 2016). There were two different Straight Outta Compton trailers delivered to the affinity groups. The trailer shown to white people boxed into their own ethnic affinity group portrayed the film as a “gangster movie” full of violence and conflicts with police. The trailer shown to black people pegged into their respective affinity group portrayed the film as a biopic, with violence contextualized as “‘protest’ imagery created by the very relatable, beloved members of N.W.A.” (Newitz, 2016). In another instance, Facebook was sued by its users over “over alleged discriminatory policies that they say violate the US Federal Housing Act of 1964” (Farivar, 2016). The ethnic affinity grouping allowed marketers to offer housing advertisements that exclude ethnic affinity groups (Angwin and Parris, Jr., 2016) [32]. Under apartheid, the South African Broadcasting Corporation (SABC) also provided different media content for its official race categories. Channel TV1 was reserved for whites and TV2/3 was reserved for blacks. Their language policy called for “culturally authentic communication” [33]. English terms like “potato chips” and “toothpaste” were prohibited by TV2/3. There were no corresponding terms in Zulu, Xhosa, or Tswana, so advertisers were forced to describe them as “slice of potato fried in oil” or “the soap that cleans your teeth” [34]. Separate media content was designed to reinforce racial segregation. While big data content delivery does not filter as sharply, discriminatory content delivery via filter bubbles threatens to segregate cultural groups, with adverse effects on oppressed groups and social cohesion. The poor black majority thus has many reasons to worry about government and commercial surveillance. When asked about collection of student data, a high level e-education official, Sipho* said, “There are a lot of contentious problems. It’s very much linked the right to privacy of individuals and so forth”. When asked about the fact that Google and the United States government will be tracking the youth from a young age, Sipho responded that “we are not having such a discussion at the moment” [35].