Essay by Evelyn Aswad
Herman G. Kaiser Chair in International Law, Professor of Law, and Director of the Center for International Business and Human Rights at the University of Oklahoma College of Law. The author wishes to thank Jason Pielemeier, Michael Samway, Sabeena Rajpal, and Rebeca West for reviewing the article. The views are solely those of the author.
A PDF of this Essay may be found here or by clicking on the title above.
On May 31, 2016, the European Commission together with Facebook, Twitter, Google’s YouTube, and Microsoft (the ICT Companies) issued a voluntary Code of Conduct on Countering Illegal Hate Speech Online that requires removal of any hate speech, as defined by the European Union.1 The impetus for this Code included the rise of intolerant speech against refugees as well as concerns that hate speech fuels terror attacks.2 Germany in particular has garnered significant media attention for its robust crackdown on Internet hate speech targeting refugees.3
The road to governmental excess can be paved with good intentions in times of crisis, with the laudable goals of refugee protection and terror prevention reflecting such good intentions in this case. Civil society groups have criticized the Code as endangering freedom of expression and lamented their exclusion from its drafting process.4 A Harvard law professor was initially outraged at the U.S. companies for selling out on First Amendment free speech principles, but ultimately decided that society shouldn’t expect private sector actors to protect speech.5
With governments no longer holding exclusive power to resolve international crises and companies playing significant roles in global affairs, it is not surprising that U.S. companies find themselves in the middle of this debate, triggering three key questions: Should the ICT Companies have agreed to this Code? Is it fair to expect that U.S. companies will seek to respect freedom of expression abroad when democratic governments raise concerns about risks to refugees and others? How should these companies handle future requests to take down hate speech in Europe?
This Essay seeks to unpack these questions. The Essay first examines the Code’s hate speech definition as well as the companies’ commitments. It then analyzes whether such commitments comport with two leading business and human rights frameworks that the U.S. Government supports: the United Nations Guiding Principles on Business and Human Rights and the Global Network Initiative. The Essay concludes that the Code is inconsistent with these frameworks as well as U.S. governmental expectations for U.S. companies operating abroad. This Essay then proposes ways forward for the key enforcers of the speech code, the ICT Companies.
I. The European Code’s Definition of Hate Speech and Related Commitments
While there is no universally accepted definition of hate speech,6 the Code defines hate speech according to the 2008 European Framework Decision 2008/913/JHA (the Framework) as “all conduct publicly inciting to violence or hatred directed against a group of persons or a member of such a group defined by reference to race, colour, religion, descent or national or ethnic origin.”7 The definition appears to contain four components: the speech must be (1) public (2) rise to the level of “incitement” (3) to violence or hatred, and (4) be directed against particular enumerated groups.8 The Framework itself further defines hate speech more broadly, including “publicly condoning, denying or grossly trivializing crimes of genocide, crimes against humanity and war crimes . . . when the conduct is carried out in a manner likely to incite to violence or hatred” against the enumerated groups.9 The breadth (and vagueness) of the scope of “hate speech” is emphasized by a provision allowing Member States to “choose to punish only conduct which is either carried out in a manner likely to disturb public order or which is threatening, abusive or insulting.”10 This provision implies the definition of hate speech is so broad that it includes speech that is not likely to affect the peace, as well as mere insults. The Code and Framework contemplate criminal sanctions for individual perpetrators and do not reference international law standards on freedom of expression.11
The Code states the ICT Companies will be “taking the lead in countering the spread of illegal hate speech online” and provides a number of specific commitments.12 For example, the ICT Companies pledge to have processes to review notifications of hate speech and public policies prohibiting “incitement to violence and hateful conduct.”13 Upon receipt of a valid notification of hate speech, the ICT companies agree to review the requests against their own guidelines and, as necessary, with national laws that implement the Framework.14 They are to review the majority of the notifications within 24 hours and remove or disable illegal content.15 The ICT Companies pledge to work with civil society partners to promote their ability to report illicit speech, to feature such “trusted reporters” on their websites, and to promote counter-narratives to the hate speech that they have suppressed.16 The European Commission and the ICT Companies will produce an assessment of this program by the end of 2016.17
II. Contemporary Business and Human Rights Standards
A. The United Nations Guiding Principles on Business and Human Rights
This summer marks the fifth anniversary of the U.N. Human Rights Council’s unanimous adoption of the Guiding Principles on Business and Human Rights (the U.N. GPs) in a resolution co-sponsored by the U.S. Government.18 The U.S. Government expects American companies to implement the U.N. GPs and treat this framework as a floor rather than a ceiling in their operations.19 The U.N. GPs set forth global expectations for how companies should act when confronting human rights challenges, such as requests by governments to censor speech. Companies should respect human rights throughout their operations, which means avoiding infringing on human rights and addressing adverse human rights impacts.20 Under the U.N. GPs, companies should conduct human rights due diligence and engage actively with external stakeholders in assessing human rights challenges.
General Principle 12 of the U.N. GPs defines the content of “human rights” according to four key U.N. instruments.21 The official commentary provides that companies may need to refer to additional U.N. instruments for guidance. The U.N. GPs do not advise (or authorize) companies to use regional or national human rights instruments to define the content of internationally recognized human rights. The U.N. GPs correctly define “human rights” by reference to international instruments rather than regional instruments, which can depart from international norms, providing fewer human rights protections at times. Where there is a conflict between local law and internationally recognized human rights, the U.N. GPs provide that companies should seek, to the extent possible, to respect international human rights while ultimately complying with local law and should address adverse human rights impacts.22
Of the four instruments highlighted in General Principle 12, the International Covenant on Civil and Political Rights (ICCPR) is the most relevant to the issue of hate speech. The ICCPR has 168 State Parties, including the United States.23 Article 19 provides for a broad right to seek and receive information of all kinds, regardless of frontiers, and through any media.24 It permits states to limit speech when a three prong test is met. To be valid, speech restrictions must be: (1) “provided by law” (i.e., properly promulgated and provide appropriate notice) and (2) “necessary” (i.e., the speech restriction must, among other things, be the least intrusive means of achieving governmental purposes) (3) to achieve an enumerated legitimate government objective (e.g., protection of the rights of others, national security, public order, public health or morals).25 Thus any limitation on speech, including hate speech, must meet Article 19’s tripartite test to be valid.
Article 20(2) of the ICCPR provides for mandatory bans on speech for “[a]ny advocacy of national, racial or religious hatred that constitutes incitement to violence, discrimination, or hostility.”26 During the ICCPR negotiations, the United States (led by Eleanor Roosevelt), the UK, and others argued against this provision as it was ambiguous and likely to be misused by dictatorships. Ultimately, however, the main proponent—the Soviet Union—succeeded in keeping the provision in.27 Not only was the negotiating history of Article 20(2) contentious, but this provision remains the subject of much controversy. For example, a 2006 U.N. report found that there was no consensus among states about the meaning of key terms in Article 20, such as “incitement,” “hatred,” and “hostility.”28 The U.N. subsequently undertook a process to convene experts in four regional workshops to propose a way forward for determining the scope and content of Article 20.29 The experts found a lack of prosecution of “real” incitement situations and pervasive prosecution of minorities under the guise of “incitement.”30 This process of dialogue among experts (but not ICCPR State Parties) culminated in the Rabat Plan of Action, as the last experts’ consultation was held in 2012 in Rabat, Morocco.31 The Rabat Plan proposes definitions for key terms in Article 20 and notes additional tools for combatting intolerance that do not involve banning speech,32 but U.N. member states have not endorsed the Rabat Plan and the scope of Article 20 remains under discussion.
While Article 20(2) contains ambiguities, there are indications that the article is to be read narrowly. For example, to trigger Article 20(2), there must be “advocacy”, i.e., the speaker must intend to incite others, and thus it does not encompass all speech that listeners find repugnant or hateful.33 Moreover, the negotiating history makes clear that Article 20(2) was included to prohibit speech that would rise to the level of creating the next Holocaust and not to capture every hateful or offensive expression.34 In particular, it was meant to capture speech that would galvanize action against the target group and not speech that merely offended the target group.35 In addition, the Human Rights Committee believes any restriction under Article 20(2) must meet Article 19’s tripartite test as well as other ICCPR requirements in order to constitute a valid restriction on speech.36
Although not specifically noted in General Principle 12 or its commentary, the U.N. Convention on the Elimination of Racial Discrimination (CERD) is also relevant. Article 4 requires that States Parties, with “due regard” to other human rights including freedom of expression, prohibit “dissemination of ideas based on racial superiority or hatred, incitement to racial discrimination, as well as . . . incitement to [racial violence].”37 The U.N. Committee charged with monitoring implementation of the CERD has recently issued its recommended interpretations of Article 4. While some of the guidance is ambiguous at times, it does provide that any racist speech restrictions must pass ICCPR Article 19’s tripartite test (e.g., to be valid, speech restrictions must give society proper notice and serve as the least intrusive means to combat racism).38
As multinational companies often seek help from their home countries when becoming entangled in human rights challenges abroad, it is important to note the U.S. approach to this international framework. Given its broad speech protections, the United States has taken a reservation to both ICCPR Article 2039 and CERD Article 4.40 On many occasions the United States has explained its reasoning. For example, in response to a U.N. survey on how countries interpret Article 20, the United States recalled President Obama’s historic remarks in pre-Arab Spring Cairo, in which he said “suppressing ideas never succeeds in making them go away.”41 The U.S. submission noted that censorship raises the profile of speech—thereby magnifying its content and authors—and also drives ideas underground, making them fester and more dangerous.42 The submission recalled early restrictive U.S. speech laws, including bans on inciting hatred against the government and censorship of criticism of slavery, which were counterproductive, as public debate and scrutiny proved a better means for progress than censorship.43 The United States now adheres to high bars for banning repugnant expression: the only hateful speech that may be banned regards advocacy that incites imminent and true threats of violence, which are statements a reasonable recipient would understand to mean the speaker intends him or her bodily harm.44 However, the submission noted that the United States does not “sit idly by” when “toxic expressions” are spreading, but rather deploys a robust array of local, state, and national governmental tools to combat intolerance without banning speech, including various forms of proactive outreach to minority groups, conflict resolution services, training programs, dialogue initiatives, and the enforcement of discrimination and hate crime laws.45 The ICT Companies can draw important cues about how their home country is likely to view rigorous Code enforcement from this background.
B. The Global Network Initiative (GNI)
Another framework the U.S. Government supports is the GNI, a multi-stakeholder initiative comprised of companies, civil society, academics, and investors to provide guidance to companies facing freedom of expression and privacy challenges in their online operations.46 The GNI company participants are Google, Yahoo!, Microsoft, Facebook, and LinkedIn (but not Twitter). As GNI members, these companies have committed to respecting freedom of expression, as defined in ICCPR Article 19 (and not in regional human rights instruments). The GNI does not explicitly reference ICCPR Article 20 or CERD Article 4 in its standards. If participating companies encounter requests that do not comport with GNI standards, they are supposed to push back to the extent possible (e.g., by challenging governments in domestic courts or seeking the assistance of international human rights bodies) before complying with local law.47
C. Comparison of the Code with International Standards
As the Code is pinned to regional concepts of illegal speech rather than international instruments, it is important to consider how European approaches to freedom of expression may differ from the international human rights regime (and therefore from the U.N. GPs and GNI). The European Convention on Human Rights provides language similar to ICCPR Article 19 for freedom of expression,48 but it has been interpreted by the European Court on Human Rights in ways that depart significantly from ICCPR interpretations involving hate speech by the U.N. Human Rights Committee. For example, the European Court has upheld French criminal sanctions for Holocaust denial without engaging in a serious analysis of whether the restriction on speech was permissible (e.g., necessary to achieve governmental aims) as it deemed the offensive speech substantively “incompatible” with the Convention and thus unworthy of scrutiny.49 The Court’s approach of removing such offensive speech from any protection of the Convention has been called the guillotine effect.50 The U.N. Human Rights Committee, on the other hand, has recently stated that the ICCPR does not condone general prohibitions on denials of historic facts.51
Similarly, the U.N. Human Rights Committee and the European Court have also approached blasphemy, or speech that offends religious sensibilities, differently. For example, the European Court upheld Austria’s decision to engage in prior censorship of a film that dealt with Christian beliefs in a highly offensive manner because it was “disparaging religious doctrines.”52 The Court held that protecting citizens from having their religious feelings insulted was a legitimate government purpose53 and that banning the film was necessary, as it could have offended the majority Catholic population and thus disturbed the peace (though it cited to no evidence in reaching this conclusion).54 The U.N. Human Rights Committee, on the other hand, has recently stated that prohibitions on lack of respect for religions or beliefs are generally incompatible with the ICCPR unless they meet the high standard in Article 20(2) and other treaty provisions such as Article 19’s tripartite test.55 From such examples, it is evident that the European human rights system’s approach to hateful speech is not always in line with the international free speech protections.
As noted, the U.N. GPs and GNI provide that companies should respect internationally recognized human rights. At a minimum, this means the ICT Companies need to assess any governmental speech restriction (even those arising from ICCPR Article 20(2) or CERD Article 4) by ICCPR Article 19’s tripartite test. In this case, the Framework and Code are inconsistent with international protections for free speech in a variety of ways. For example, as noted previously, the definition of hate speech is remarkably broad and vague, apparently encompassing speech that is merely insulting, speech that is not likely to affect public order, and speech that denies historic facts.56 In addition, the Code and Framework require criminal penalties for such “hate speech.” This presumption of the appropriateness of criminal sanctions for such a broad definition of hate speech is not consistent with the ICCPR’s requirement that the ban on speech and the ensuing sanction must be the least intrusive means of achieving objectives.57 If the objectives can be achieved without bans or criminal penalties, then the bans and/or punishments are inconsistent with the ICCPR. In addition, ICCPR Article 26 requires that laws not discriminate among groups. Clearly the Code’s prohibition on hate speech only protects certain groups (primarily relating to race and religion) to the exclusion of others, which could also call into question such laws under the ICCPR’s equal protection provision. In sum, by agreeing to regional (European) rather than international standards, the ICT Companies have departed from the U.N. GPs and GNI frameworks and risk contributing to international human rights violations.58
With this background and analysis, the answers to the questions at the beginning of this Essay become fairly evident. The answer to the first question (should the ICT Companies have agreed to this Code?) is no. U.S. companies should not agree to voluntary codes of conduct on human rights matters that are not expressly linked to relevant U.N. human rights instruments. Pinning their commitments on regional and national—rather than international—human rights standards puts companies at risk with respect to global expectations embodied in the U.N. GPs as well as U.S. government expectations. Moreover, agreeing to a European code that is not linked to international standards will make it more difficult for those same companies to resist signing on to voluntary codes of conduct linked to local norms that other (less democratic) regional bodies or national governments may request in the future.
With regard to the second question (is it fair to expect U.S. companies to respect freedom of expression abroad?), the answer is yes. Under contemporary business and human rights frameworks that the U.S. Government supports—the U.N. GPs and the GNI—businesses are expected to respect human rights, as defined in U.N. instruments, in their operations, including their terms of service. They are expected to engage in human rights due diligence, work with external stakeholders in identifying and dealing with human rights challenges, and use good faith efforts to avoid infringing human rights, as defined in U.N.—not regional or local—instruments.
The answer to the third question (how should these companies handle future requests to take down hate speech in Europe?) is more complicated. Ideally, the companies would renegotiate or withdraw from the Code. Assuming there is too much water under the bridge for that to happen, the companies should make clear that they will interpret the Code in light of the U.N. GPs, which represents the global framework that both EU states and the United States support. Thus the ICT Companies should develop operating policies that are pegged to U.N. instruments, which means (at a minimum) examining take down requests against the requirements of ICCPR Article 19’s tripartite test, including whether censoring speech constitutes the least intrusive means for achieving legitimate ends. When there is ambiguity in international standards on free speech (such as what constitutes “incitement”)59 or other opportunities to promote broad expression protections, I would encourage U.S. companies as a matter of policy to interpret the ambiguity or seize other opportunities in favor of U.S. free speech approaches60 to the extent possible, given the dangers inherent in governmental censorship and the possibility of using tools other than speech bans to promote tolerance.
As provided for in the U.N. GPs and GNI, the companies should pursue all possible avenues to narrow governmental requests that restrict speech in a manner incompatible with international standards (including challenging take down requests in court), rather than abiding by a Code that is untethered to international standards. Even if the European Commission is not amenable to including civil society in discussions about the Code, companies should include civil society in their own deliberations about the Code, as provided in the U.N. GPs. If ultimately compelled to follow European or national law, then according to the GPs and GNI, companies should do so after having engaged in the type of proactive diligence outlined above and should address negative impacts on internationally recognized human rights.
In sum, agreeing to governmental requests for voluntary adherence to regional or national speech rules is not appropriate in 2016. Rather, the ICT Companies should pin their policies to the U.N. GPs and GNI and be guided by international standards when operating abroad. While the governmental concerns prompting the Code are legitimate, departing from the international human rights law framework will inevitably trigger the law of unintended consequences, in this case repeated negative impacts on freedom of expression and ultimately the weakening of those democracies, which will only undermine the noble goals of refugee protection and the prevention of terrorism.