One Law to Rule Them All

We’re Skeptical of UN-sponsored Guidance for Content Regulation

Jessica Fjeld and Sol Bensadon

An arm of the United Nations has taken up the challenges of content moderation, aiming for a September 2023 release of guidelines to help countries regulate online platforms while protecting free expression. A great idea, in principle, but the devil is in the details — or in this case, the devil is in the implementation of vague language about “protecting democracy” that’s as likely to be weaponized against dissidents in some contexts as it is to curtail election misinformation in others.

UNESCO was created in 1945 with the mission of contributing to peace and security by promoting “international cooperation in education, sciences, culture, communication and information” and “knowledge sharing and the free flow of ideas, to accelerate mutual understanding and a more perfect knowledge of each other’s lives”. As part of this mission, UNESCO is now engaging in consultations with multiple stakeholders with the objective of developing global guidelines for the regulation of digital platforms. UNESCO has already circulated two drafts and the second one of these was discussed at the Internet for Trust conference that took place last month. UNESCO has nearly 200 member states, representing a great diversity of cultures, ideas, and governance models. They also represent the full spectrum of internet freedom — both the lowest-ranked (China) and the highest-ranked (Iceland) countries in the most recent Freedom House Freedom on the Net report are members.

The law and jurisprudence around free expression are characterized by basic tensions, which can never be resolved, but are managed through the three-part test of legality, proportionality, and necessity. It’s not an area where bright lines can be easily drawn. Most basically in the context of content moderation, protections for free expression require both leaving content up (to allow a range of views) and taking content down (to ensure an environment that does not drive away marginalized folks).

This comes to a very fine point when we consider that the aim of the UNESCO guidance: “to support the development and implementation of regulatory processes that guarantee freedom of expression and access to information while dealing with content that is illegal and content that risks significant harm to democracy and the enjoyment of human rights.” These categories simply are not coherent across the range of UNESCO’s member states, and it’s not clear that there is one policy framework that could achieve them and accord with International Human Rights Law (“IHRL”).

There is a large risk that the guidance document will be misused by authoritarian states to legitimize existing arbitrary restrictions on freedom of expression and to justify new ones — and we’re not the only ones to make this point. Consider a few examples, all UNESCO member countries:

• In 2020, Nicaragua passed the “Special Cybercrimes Law,” prescribing imprisonment for whoever disseminated “fake news” online that risked “economic stability, public health, national sovereignty or law and order.” This regulation, also known as the “gag law,” was used as a tool to silence dissenters. The gag law was first applied in January 2022, after the government had already imprisoned dozens of opposition members based on their social media activity in the context of the November 2021 general election.
• In Vietnam, internet content advocating for human rights and democracy is actively censored. In other words, content that promotes “democracy and the enjoyment of human rights” is illegal. Under the 2018 Cybersecurity law, intermediaries are forced to regulate content in cooperation with the state, preventing the sharing of information that opposes the republic, threatens national security and the social order, or defies national traditions, among other broadly worded provisions” and deleting information after government requests.
• Finally, how the government of the Philippines has treated Maria Ressa — a keynote speaker at the UNESCO conference — further illustrates the use of the legal system and the spread of disinformation online to silence criticism. There were at least ten criminal cases filed against Maria Ressa, three of which were under the Cybercrime law, which has been criticized as threatening free speech.

In authoritarian states, the coherence between national law and IHRL is extremely limited: the imposition of an IHRL-based framework to produce the result of limiting domestically illegal content may well distort that framework beyond recognition. The current UNESCO guidance does not give us any insight into the application of the proposed schema in contexts where the state itself is the preeminent actor in the repression of free expression. It is not enough to simply note that the application of the International Covenant on Civil and Political Rights (“ICCPR”) the guidance document does not impact the ICCPR’s enforceability in these states and is highly unlikely to make changes on the ground. Moreover, even so-called “advanced democracies” have adopted or are considering restrictions on internet speech that many digital rights advocates find troubling — the UK’s Online Harms Bill is a good example.

Fortunately, there’s a way to avoid the worst manipulations of UNESCO’s work: require procedural guarantees, which would ensure that they can’t be used to whitewash repressive laws. The current draft already includes some such elements, such as the opposition to hostage laws2 and upload filters3 and the support for broad media literacy investments from both states and companies. However, we think additional, well-tailored measures will be needed.

A guarantee of anonymity is an essential protection for individuals’ free expression, especially but not only in repressive contexts. As the United Nations suggested, anonymity provides individuals “with a zone of privacy online to hold opinions and exercise freedom of expression without arbitrary and unlawful interference or attacks.” In addition, the guidelines should include robust whistleblower protections, mandating states to ensure that government and private-sector whistleblowers are protected from any form of retaliation, and to create internal safe channels for employees to report concerns and potential harmful choices. This is essential, considering the important role that whistleblowers have in revealing biased enforcement of the law and general misuse of regulation to restrict freedom of expression.

Finally, under IHRL, restrictions on freedom of expression need to be sufficiently precise to enable individuals to regulate their conduct accordingly. Were any state to adopt the broad and overbroad language of the guidelines, it would lack precision, as it is not clear, outside of certain edge cases, what constitutes “content that risks significant harm to democracy.” Especially in the context of authoritarian or non-democratic states, regulations with vague and overbroad language can be used to censor dissidents.

Careful attention must be paid to the language of the procedural protections incorporated into the guidelines. A good illustration of this point is transparency requirements: the same processes for review of corporate policy and decision-making that can allow countries with relatively free information environments to promote free expression, can be used in less free environments to tighten an authoritarian hold. Berkman Klein’s Lumen Database has shed light on the global climate of online expression, censorship, and rights enforcement for 20 years. Our Lumen colleagues have thought long and hard about designing transparency for the possibility of misuse and abuse, and have been consulted on the design of the new database contemplated by the EU Digital Services Act.

Sensitive data, such as personally identifiable information or materials that a state might seek to persecute groups or individuals, should still be considered for inclusion in a transparency data corpus, and made available for study, with increased security, oversight, and control. In more open contexts, this can be achieved with levels of restricted access, allowing only accredited researchers, who agree to be bound by relevant strictures, to see data that has a high risk of misuse. It would take some creative thinking to establish a data governance plan that would facilitate research where the state itself is part of the threat model.

If UNESCO is serious about protecting — and not further threatening — free expression in platform regulation around the world, the range of stakeholders who are invested in internet freedom that were represented at February’s convening was a step in the right direction, but not a destination in itself. Between now and the guidelines’ launch, the organization should have a laser focus on the thorny challenges of defining procedural protections that will be effective even in the least protective regulatory environments.