Legislative responses

Legislative or legal regulatory measures concern measures that have a clear legal basis and hence are more transparent. Unlike extra-legal measures, this set of measures is either explicitly mandated in law or is a given piece of legislation that is often invoked to justify the measure. There are two categories of legislative or legal regulatory responses to disinformation in Ethiopia. One is the prosecution of suspects under the newly adopted disinformation legislation.Footnote ⁸ Before the adoption of this relatively tailored law, prosecution of similar offences was carried out under the Criminal Code.Footnote ⁹ The second legislative regulatory measure took the form of an institutional response by which “cyber warriors” would be trained to attend to disinformation, among other hybrid cyber threats, online.Footnote ¹⁰ Until the institution tasked to train these warriors was abolished in mid-2019, this regulatory measure was duly mandated by law.

Criminal prosecution

For over a decade, the Ethiopian government has prosecuted a number of individuals, including journalists, for disseminating falsehood under a controversial provision in the Criminal Code of 2004. Article 486 of the Code criminalizes two sets of acts: (a) spreading or initiating false rumours, suspicions or false accusations of the government or public authorities and hence twisting public opinion or creating public disturbance; (b) fomenting dissent, hatred, violence or political, racial or religious disturbance by spreading accusations. This provision had long come under attack from different corners for being used to suppress legitimate dissent and freedom of the press.Footnote ¹¹ With the adoption of the new disinformation legislation, namely Proclamation No. 1185/2020, this particular article of the Criminal Code has been repealed.Footnote ¹² The adoption of this legislation has, however, not been without criticism. Beyond the concerns expressed by human rights advocacy groups that the law could be used to suppress critical voices,Footnote ¹³ 23 parliamentarians voted against the adoption of the law as well.Footnote ¹⁴ No legislation in Ethiopia's recent past has spurred such opposition in the legislature.

This legislation now constitutes the prime legislative regulatory response to disinformation in Ethiopia. Article 3 of the law provides that its main objective is to ensure “individuals will not engage in speech that incites violence, is likely to cause public disturbance or promotes hatred and discrimination against a person or an identifiable group” (emphasis added). This provision suggests Ethiopia's regulatory response focuses on addressing disinformation that presents dangers to national security and social harmony. The national security rationale common in certain jurisdictions is explicitly recognized in this legislation. This further suggests that guarding the integrity of electoral processes and democratic institutions is not a stated legislative goal of Ethiopia's disinformation law.

Of course, one of the stated objectives of the law is to “strengthen democratic governance” in the country, but this objective is not framed in a manner envisaged in the laws of several other countries.Footnote ¹⁵ That is, the objective is not to foster democracy by curbing the risk of disinformation to democratic processes and institutions, but rather to strengthen democratic governance by criminalizing disinformation (and also hate speech). And it is probable that this objective concerns the hate speech aspect of the law because it seeks to strengthen democracy by “promoting tolerance, civil discourse and dialogue”.Footnote ¹⁶ Further reinforcing Ethiopia's national security approach to disinformation is the Preamble of the law in which its underlying objectives are provided. Paragraph 2 of the Preamble states: “Cognizant of the threat hate speech and disinformation pose to social harmony, political stability, national unity, human dignity, diversity and equality”.

A closer examination of the law reveals two key lapses that undermine its effectiveness in tackling disinformation through criminal prosecution.Footnote ¹⁷ First, the law exhibits a number of substantive shortcomings; in particular, it suffers from a number of ambiguities. The ambiguity in the law starts with the term “disinformation”. Article 2(3) of the Proclamation defines “disinformation” as “speech that is false, is disseminated by a person who knew or should reasonably have known the falsity of the information and is highly likely to cause public disturbance, riot, violence or conflict” (emphasis added). This definition seems to be based on three incorrect assumptions. The first is that “falsity” has a fixed meaning, but that is not always the case as an information or idea could be only partially false, contextually false or visually false. The second assumption is that only categorically “false” information poses a problem (for example disturbance or violence) and is hence worthy of regulation. But the spread of information that is speculative but hard to verify, or information that is not entirely factually inaccurate but misleading, could be a source of destructive incidents on the ground. Finally, the provision has its sights set only on disinformation that is likely to result in physical consequences in real life: “public disturbance, riot, violence or conflict”. As such, it therefore overlooks other less tangible consequences of disinformation, such as misleading voters, swaying the course of elections or casting doubt on the validity of election outcomes. This provides additional evidence that Ethiopia's approach falls largely in the approach to disinformation regulation common in countries where national security is the prime driving factor.

The term “dissemination”, which forms a key part of the crime of disinformation, is also ill-defined. It is defined as “spreading or sharing [sic] a speech on any means for many persons [sic] but it does not include like or tag on social media”.Footnote ¹⁸ This provision must be read together with the provision that criminalizes disinformation and article 2(3) that defines the notion of “disinformation”. The crime of disinformation is formulated as follows: “disseminating [sic] any disinformation on public meeting by means of broadcasting, print or social media using text, image, audio or video is a prohibited act”.Footnote ¹⁹ The way in which “disinformation” is defined, as outlined above, suggests that criminal liability would fall solely on the person who disseminated the information while knowing, or who ought to have known, the dire consequences it would cause. This reading of the provision may restrict criminal liability to the person who originally shared the disinformation, but not to those who further share or circulate the content. But the definition of “dissemination” appears to broaden the scope of criminal liability. Except for the specifically mentioned acts of “like or tag”, any act of spreading or sharing any disinforming content would therefore attract criminal liability.

One notes at this point that the exceptions are not only exhaustive but also seem to have in mind interfaces in currently popular social media sites such as Facebook. This parochial vision of social networking makes it lack the key principle of technological neutrality in internet policymaking by basing the normative standard on the technology of the time. This principle of internet regulation holds that regulation should be “flexible, time-proof and open to technological change”.Footnote ²⁰ What would happen when new technological interfaces that facilitate easier spreading of online content emerge? Or does the provision even, for example, exclude retweets on Twitter, a popular social networking site? It is also vital to note that the exceptions do not concern other channels of dissemination such as print or broadcasting. What if a particular television show airs an excerpt with some potentially problematic content? The law leaves such questions open.

Another form of ambiguity is evident in the significant variations between the Amharic and the English versions of the law. This point is worth noting because when such variations occur, the Amharic version would prevail. But still, laws in Ethiopia are often first drafted in English and then translated into Amharic, mainly because most Ethiopian laws are transplanted from elsewhere. While it is not known in what language the law was originally drafted, the divergence might cause the true intent of the lawmakers (and drafters) to be lost in translation. For instance, the Amharic version of the provision defining disinformation – i.e. article 2(3) – embodies a condition not specifically mentioned in the English version. According to the Amharic version, “disinformation” is false speech disseminated by a person who has not taken reasonable steps under the circumstances to ensure the veracity of the information. This is neither patent nor latent in the English version. The implication of such divergence is that if courts were to rely on the Amharic version only, that version would be more protective of free expression than the English version.

The law borrows a notion from the 2018 Egyptian media law that equates social media accounts that have more than 5,000 followers to media establishments.Footnote ²¹ But Ethiopian law uses this numeric threshold of followers as an aggravating circumstance, i.e., that when disinformation is spread by print, broadcast media or social media accounts with more than 5,000 followers, the applicable penalties would increase from jail time of maximum one year to maximum three years or fines from 50,000 ETB to 100,00 ETB. Two problems, however, inhere in this proviso: the numeric threshold appears to be arbitrary; it also does not clearly indicate whether the relevant account should relate to those owned by individuals only or also to groups, associations etc. as well. If the latter is the case, it is not clear how this would play out in terms of criminal liability. In particular, the nature of criminal liability, as well as who, when it comes to groups or associations, would be held liable, and in what form, is also far from clear.

What is ironic about the new disinformation law is that not only does it tend to legitimize problematic speech under the name of “exemptions”, thereby undermining its legislative objective, but also the exemptions under the law are framed somewhat broadly as well. Article 6 of the Proclamation states that speech would not be considered “disinformation” when it is part of (a) academic study or scientific inquiry; (b) a news report, analysis or political critique; (c) artistic creativity, performance or other form of expression; (d) religious teaching; or (e) where the speech has been made after the speaker took reasonable efforts in the circumstances to check the veracity of the information. While exemptions are definitely required to uphold free speech, the exceptions must be carefully defined and be qualified. This is more so in the current highly polarized and ethnicized political environment in Ethiopia. Otherwise, under the banner of, for instance, academic study or scientific inquiry, deeply problematic speech could easily escape restriction. The other exemptions are also problematic. Dangerous and consequential ideas and information may be conveyed in the name of artistic or journalistic ventures or even religious preaching. For instance, the exception for academic or scientific inquiry must be qualified to allow only potentially disinforming content that is expressed for “purely” academic study or scientific inquiry.

The second lapse is that the law faces implementation challenges partly because of the way in which it envisions institutional roles and the nature of the problem. In some respects, the law appears to envision a co-regulatory solution to the problem of disinformation. First, it generally obliges social media companies to put in place internal rules that govern the notice-and-takedown of problematic speech on their platforms.Footnote ²² Secondly, it requires these companies to “endeavour to suppress and prevent dissemination” of problematic speech on their platforms.Footnote ²³ In a way, this proviso further solidifies and legitimizes the enormous power that social media companies wield in content moderation. However, the law does not envision any penalties against those companies failing to comply with those sets of obligations. Should a secondary legislation be later enacted (as envisioned in article 8(7)), it may prove problematic.

On pain of being penalized, these corporations would, more often than not, resort to aggressive content moderation measures, thereby significantly curtailing free speech on their platforms. But, of course, this would most likely apply to future local social media companies – if they come into the picture at some point – and not global giants like Facebook and Twitter. As a poor country, Ethiopia has no meaningful leverage by which to force companies such as these, which generally have a smaller user-base in the country.Footnote ²⁴ In the current reality, that undercuts the potential of the proviso in which American-based social media companies like Facebook are the prime spaces for disinformation affecting Ethiopia. Thirdly, it requires these companies to remove impugned content within 24 hours of receiving notification.Footnote ²⁵ However the law leaves it open as to who is entitled to submit notification – is it anyone, or is it the police or the other bodies mentioned in the law, i.e. the Ethiopian Human Rights Commission and the Ethiopian Broadcasting Authority (EBA)?

The assignment of institutional roles under the law is also far from clear. The nation's media regulator is tasked to carry out two functions. First, it shall compile and disclose a report detailing the degree to which social media companies comply with the lawFootnote ²⁶ (and the purpose of this report appears to be simply “notifying” the public). Such a duty of reporting on media regulators, in the context of disinformation regulation, is not new. French misinformation law also requires the media regulator to compile similar reports,Footnote ²⁷ but it does not stop there as, based on that report, measures could be taken against delinquent parties. Where the delinquent is a media agency, for example the regulator may revoke its broadcasting licence. In cases where delinquents are other actors, the report provides the basis for action by other regulators. But as the law currently stands, Ethiopia's disinformation law says little regarding the goal of the reporting. The stated goal of “notifying the public” serves little purpose in terms of countering the problem of disinformation.

The second task of the EBA is to run media literacy and public awareness campaigns.Footnote ²⁸ While this provision rightly advances a widely accepted strategy in dealing with disinformation, it leaves many questions open. One is that it is not entirely clear why and how the EBA is the pertinent body to raise public awareness. As a regulator of media houses, it is best positioned just to follow the regular regulatory channels in order to tame rogue media organizations. A series of regulatory tools are in place to address media organizations’ conduct in the context of disinformation and/or hate speech. When it comes to media literacy, the global approach varies, but an emerging trend is that media literacy should not be delegated to a single institution, but instead to several organs within their sphere of influence. Another growing trend in the long-term approach to media literacy is to make it part of school curricula. The US states of California and Massachusetts have, for instance, taken steps towards introducing media literacy lessons in public schools.Footnote ²⁹ Ethiopia's approach towards media literacy should be reoriented along these lines.

The Ethiopian Human Rights Commission also assumes a role under the disinformation legislation as the Commission is tasked to raise public awareness regarding hate speech.Footnote ³⁰ Under its founding law, the Commission is charged to raise public awareness about human rights and also conduct human rights education.Footnote ³¹ This suggests that its additional function under the disinformation law is based on the assumption that “hate speech” – as opposed to “disinformation” – is a human rights problem and that the Commission is best placed to raise public awareness. As noted above, the EBA is tasked to raise public awareness regarding disinformation, but disinformation is as much a human rights problem as is hate speech, and disinforming attacks could well be a source of egregious human rights violations similar to hate speech. This regulatory conception of “disinformation” is problematic because it encourages cumbersome responses against acts of online disinformation, thereby burdening human rights, and in particular free expression. But the law also leaves out other pertinent authorities from the purview of the law. Entities like the Information Network Security Agency (INSA), which is given key roles in cybercrime investigation and prosecution, should also have been tasked under the law.

So far, the law has been used to prosecute disinformation in only a handful of cases. In a recent alleged “fake news” case, the police reportedly invoked the anti-terrorism legislation – as opposed to the new disinformation law – to prosecute a freelance journalist named Yayesew Shimelis.Footnote ³² After some confusion as to whether he was being charged under the anti-terrorism law or the disinformation proclamation, the journalist was charged, but later bailed, under the newly adopted disinformation law.Footnote ³³ He is the first individual to be charged under the disinformation law for disseminating disinformation relating to Covid-19 through his social media accounts, but the content leading to his charge had been posted days before the new law entered into force on 23 May 2020. Hence, the law in this case has been used retroactively, thereby violating the non-retroactivity of criminal law under the Ethiopian Constitution.Footnote ³⁴ Elisabeth Kebede, a volunteer lawyer at the Ethiopian Women's Lawyers Association, has been charged by the Harari Regional State Police under the new law for circulating disinformation regarding Covid-19.Footnote ³⁵ The current status of her case is unknown, however in early May she was released from detention on bail. The latest defendant under the disinformation law is Temesgen Desalegn, managing editor of a private magazine, who thus far has been briefly arrested, but not formally charged.Footnote ³⁶ As these early cases suggest, the disinformation legislation is being used against journalists and dissidents who voice their criticism, thereby clouding its effectiveness in addressing the real problem of disinformation.

Militarization of social media

Deploying “warriors” on social media platforms is the other regulatory response introduced through legislation. The initial idea of trained personnel who would debunk falsehoods on social media sites was put forward by the then governing party in 2017, the Ethiopian Peoples’ Revolutionary Democratic Front (EPRDF).Footnote ³⁷ Citing the proliferation of disinformation and hate speech online, particularly targeting the party, the ideological outlet of the Front stressed the imperatives of training and deploying a social media army. Shortly after, the idea took a different direction, and the government introduced legislation that created a “Cyber Army Development Institute” within the nation's cyber command, INSA.Footnote ³⁸ The law envisioned the Institute as a government higher education institution that would, among other things, train personnel needed for leading the nation's cyber force.Footnote ³⁹ A key objective of the Institute was provided under article 4(1):“To provide regular education to key role players in the development of cyber force and develop disciplined cyber force capable to protect national interest and respond to the needs of the country's cyber warfare”.

In a stark departure from the initial idea, the law neither explicitly mentioned a “social media army” nor mentioned problems of hate speech and disinformation. The Council of Ministers, the federal cabinet, recently abolished the Institute, and without any clear rationale,Footnote ⁴⁰ but it is interesting to note that the regulation abolishing the Institute transfers the powers and functions of the Institute to INSA.Footnote ⁴¹ This suggests that INSA would still retain the role of training personnel for the nation's cyber army that would combat, inter alia, disinformation, however little is known about whether this has been the case in reality. Also unclear is whether the Institute had trained cyber warriors before it was disbanded and whether its graduates are at work, for example within INSA. INSA is empowered under its establishment law to “take all necessary counter-measures to defend any cyber or electronic attacks on information and computer-based infrastructures or systems or on citizens’ psychology”.Footnote ⁴² One might wonder if such countermeasures, which INSA itself – as shall be noted below – has claimed to have taken to counteract cyber-attacks, have been taken to deal with social media disinformation. What is clear, however, is that the idea of a social media army does not seem to have been abandoned altogether. Reinforcing this point is the fact that shortly before the Cyber Army Institute was abandoned, the government tabled a Bill in late 2018 that would create a cyber force as well as naval and space forces.Footnote ⁴³ The Bill is yet to be adopted by the legislature, but it signalled that the government had not abandoned the idea of a cyber army, including a social media army. It is still not entirely clear, however, if the envisaged cyber force would have any role in dealing with disinformation, including foreign information influence operations or attacks.

But, the idea of creating a social media army was in itself unlikely to be problematic. That the idea was pursued openly, and with a clear legal basis, was promising. Such transparency would create a sense of certainty, and hopefully accountability. However, the initiative should have been taken at the national level and pursued more vigorously. Such efforts would have allowed the government to look to long-term measures in dealing not just with disinformation but also with all other forms of hybrid threats presented by the internet. Of course, the idea of a social media army has some semblance to the practice in a number of authoritarian countries. Countries like China – known for its fierce online censorship – have been suspected of secretly hiring about two million individuals to weigh in on social media debates on behalf of the government.Footnote ⁴⁴ But these government employees do not appear to engage in constructive debates, but instead infuse social media sites with information that praises the government and its achievements, thereby overwhelming users engaged in other useful debates.Footnote ⁴⁵

The Ethiopian government must distance itself from such unhelpful engagements as this would further distort the flow of information, burden legitimate speech, and ultimately reinforce fake news. The aim should be to counter falsehood by presenting the public with accurate and timely information based on reliable and verifiable facts, rather than dictate the truth. Had it not been aborted, the idea would have fulfilled some sort of useful purpose in dealing with disinformation, at least in terms of the kind of role Sweden bestowed on its “Psychological Defence Authority”.Footnote ⁴⁶ Sweden created this body to counter false and misleading information by presenting factual information to the public. As such, it does not engage in further offensive measures, or what is otherwise called “persistent engagement”.

Non-legislative responses

Non-legislative measures are those with no clear legal basis and bereft of any transparency or accountability as well. Internet shutdowns are one such commonly taken measure against the circulation of disinformation online in Ethiopia. Web filtering, such as blocking websites, is the other form of non-legislative regulatory measure used in Ethiopia, but it has become less common in the past two years.Footnote ⁴⁷ With the apparent regression in political reforms in the country, there are, however, signals of a return to past practices of web filtering. A growing number of websites and apps hosting content critical of the government are now inaccessible in Ethiopia.Footnote ⁴⁸ Internet shutdown is a more recent tool of “informal” content regulation.Footnote ⁴⁹ Since the widespread protest against the government in 2015 – in the orchestration of which social media platforms played a central role, the government resorted to internet shutdowns on a number of occasions. Be it to stop the spread of leaked text of national university entrance examinations or allegations of brutal crackdown on protestors, the internet has been shut down by the government several times in the past few years.

Fighting disinformation is the more recent and persistent rationale invoked by the government when shutting down or throttling internet use in some parts of the country.Footnote ⁵⁰ But the government rarely offered a legal justification for resorting to shutdowns.Footnote ⁵¹ Often, the government presents unpersuasive defences for its opaque shutdown practices. One may recall here the claim that since the internet is neither water nor air, it may therefore be shut when the government deems there is a threat to national security.Footnote ⁵² But more recently, the Office of the Federal Attorney General, in its formal comments on a report of the UN Special Rapporteur on Freedom of Opinion and Expression that lamented government shutdown practices, coincidentally offered a legal justification of sorts for internet shutdowns.Footnote ⁵³ This casual legal justification has two prongs, one of which is that the Office notes that the 2013 law that re-established INSA empowers the Agency to “keep the country safe from any threats against national security and [said] it can take measures when the necessity arises”.Footnote ⁵⁴ According to the Office, this law provides a legal basis for internet shutdowns.

A closer look at this law does not, however, support this claim. Nowhere in this law is the Agency's power to cut internet access for national security purposes either explicitly stated or implied. The closest this law comes to stating it is when it empowers INSA to take “countermeasures” for cyber-attacks against the nation's critical infrastructures, like power grids and “citizens’ psychology”.Footnote ⁵⁵ As the nation's cyber command, this proviso simply empowers INSA to take counter-measures – otherwise called “strike-backs” – against perpetrators of cyber-attacks. Strike-backs, therefore, are a sort of in-kind retaliation by resorting to proportionate cyber-attacks. If the attack was, for instance, in the form of a Distributed Denial of Services (DDoS) attack, the Agency may retaliate with similar DDoS attacks or large-scale spreading of malware. This means two things: first, when cutting internet access, INSA would not be engaging in the lawful strike-backs envisioned in its establishment law. Instead, it would simply be unlawfully restricting the individual right to free expression, including the right to access information online. Second, most trigger factors for internet shutdown in Ethiopia, such as social media disinformation, do not really fall under the rubric of cyber-attacks imagined in the law. Disinformation is a form of hybrid cyber threat, but not quite a cyber-attack warranting lawful strike-backs.Footnote ⁵⁶ Even if it were, how would a state hit back proportionately against coordinated disinformation campaigns of non-state actors? INSA had previously invoked a similar line of reasoning to justify the brief internet blackout in December 2019.Footnote ⁵⁷ At the time, INSA claimed the brief shutdown was needed to fight off active cyber-attacks against the nation, but this justification lacks any legal basis. The second part of the Federal Attorney's legal justification is that internet shutdowns are seldom taken with maximum restraint, and in line with international human rights and national constitutional standards of “legality, legitimacy, necessity and proportionality”.Footnote ⁵⁸ Not only are internet shutdowns occurring frequently – and, at times, for extended periods – but they also lack clear legal basis, thus failing to meet the requirements of “legality”. This makes it immaterial to consider whether the other requirements of necessity and proportionality are fulfilled.

Overall, recurrent episodes of arbitrary internet shutdowns in Ethiopia suggest that the government saw value in internet shutdown as a useful regulatory measure. But it could have grounded such shutdowns on a sound legal basis in the recent cybercrime law.Footnote ⁵⁹ This law potentially offers a defensible legal basis for gateway measures like internet shutdown. Moreover, because the law requires a prior court warrant before such measures are taken, it also envisages safeguards against abuse. The recurrence of network disruptions, and their impact on the enjoyment of human rights, would be mitigated when each request for an internet shutdown had to be considered by the courts in light of national and international human rights standards.

Electoral law

Ethiopia's omnibus electoral law of 2019 has some provisions that do tend to address election disinformation. For instance, it prohibits engaging in “activities that in any way disrupt, destroy or frustrate the election campaign of other parties or cooperate in such acts”.Footnote ⁶⁵ It similarly prohibits dissemination of false information, claiming that the secrecy of the ballot has not been maintained.Footnote ⁶⁶ These prohibitions appear to cover common election-related disinformation. But the scope of the law's overall orientation is directed at duly registered political parties, their candidates, members, supporters and agents.Footnote ⁶⁷ As such, it envisions threats of disinformation only from domestic actors, specifically from registered political actors. This means that it would not ordinarily apply to those who engage in those impugned acts without any political registration or affiliation. Foreign state and non-state actors that engage in those acts would not therefore be regulated by the law. This problem of scope makes this new law slightly outdated in light of modern-day disinformation perpetrated across borders through the help of modern communication technologies.

To an extent, the electoral legislation addresses the key issue of campaign finance. First, it imposes a series of regular reporting, bookkeeping and auditing obligations on political parties regarding their sources of income and donations.Footnote ⁶⁸ Second, it bans donations from certain actors, particularly foreign governments and organizations as well as non-governmental bodies.Footnote ⁶⁹ Such constraints on sources of income help fend off not only influence on political parties but also interference in the electoral process. But these parts of the law overlook a key part of recent electoral laws in the digital age: regulation of political advertisements online. Much of the election-related disinformation in other countries is undertaken through the use of purchased political advertisements by foreign as well as domestic actors. Recent electoral law reforms, for instance in the US and Canada, introduce a series of transparency obligations on social media platforms to disclose the identity of those purchasing political adverts and the attendant price, among other details.Footnote ⁷⁰ Corollary to such duties is a system in some countries by which political parties and candidates pre-certify social media adverts by election regulators.Footnote ⁷¹ This means that major social media sites would not allow adverts that have not been pre-certified by the appropriate electoral body.

In spite of its overall comprehensive scope, Ethiopia's electoral legislation leaves much-needed regulatory details to secondary legislation. It relegates to a series of directives to be adopted by the National Electoral Board of Ethiopia (NEBE) the responsibility to hammer out a number of issues, such as the manner of conducting election campaigns and media code of conduct on election coverage,Footnote ⁷² but no such directive that regulates the issue of social media mediated disinformation has been made public. In light of the law's overall narrow scope in regard to election-related disinformation, it is less likely that upcoming directives would be able to directly address the issue.

A related limitation of recent electoral laws is that they appear to underestimate the role of social media in shaping the course of elections. Although adopted as recently as 2019, the Proclamation has within its purview only legacy media as the prime medium of election-related communications. Nowhere in this law is the role of the new media taken into consideration in regulating election-related information dissemination and communication. Mass media, which primarily concerns print and broadcast media, both of which are within the regulatory reach of the government, appears to be the law's sole focus.Footnote ⁷³ It is only remotely that some provisions of the law actually touch on the sort of election disinformation threatening elections. One such provision, for instance, prohibits “abuse of position of power”, by which any political party may not use its position of power to influence – which includes “special opportunity or influencing ability” – to bribe or use any other illegal methods to further its interests.Footnote ⁷⁴ This broad framing might capture the use of social media influence. Similarly, article 136 prohibits “corrupt practices”, including using any inducement to sway the opinion of voters, but both of these rules apply only to political parties registered under Ethiopian law.

Overall, Ethiopia's electoral law has its sights set on registered political parties, candidates and their supporters. But, importantly, the law generally overlooks the role of technology companies in altering the course of elections, particularly by offering a platform for actors (state or non-state, domestic, regional or international) to manipulate voting outcomes through various methods, including disinformation. Assisted by the ever-evolving and sophisticated technological environment, the world of digital political campaigning is deeply complex, involving invisible actors and practices. This is further complicated by the transnational nature of the network on which it operates. A layer of technology companies operates in the ecosystem: from data brokers, data analytics firms, campaign platforms to AdTech and social media networks. So, behind the curtain of social media platforms from where user data are harvested and later used to target those users with political adverts, a number of invisible and, hence, unaccountable players exist.

The Cambridge Analytica scandal that plagued Facebook a few years ago is a good case in point.Footnote ⁷⁵ This data analytics company used unlawfully obtained personal data of about 87 million Facebook users in order to engage in disinformation campaigns during the US presidential elections. And, of course, the scandal hit close to home as well when similar allegations of disinformation campaigns occurred during the presidential elections in Kenya (2013, 2017) and Nigeria (2015).Footnote ⁷⁶ Unlike the laws in India, for instance, Ethiopian law does not envision social media platforms as key stakeholders in the fair and peaceful conducting of elections. It does not go beyond imposing a very general “duty to cooperate” on everyone in the proper implementation of its provisos.Footnote ⁷⁷ With increasing use of the internet in Ethiopia, and the rising role of social media in socio-political discourse, there is an urgent need to address the role of social media platforms generally, and during the election cycle in particular.

The NEBE is reportedly drafting a Directive that would partly address the issue of election-related disinformation in the social media environment. This draft Directive reportedly addresses two themes.Footnote ⁷⁸ First, it seeks to regulate the use of social media by political parties and candidates for political campaigning purposes. As such, the object of this Bill seems to be outlining a code of conduct for these political actors when using social media platforms during the election season. Second, it will also set out duties on political parties to provide ethical education to voters, particularly their supporters and constituency. Once adopted, this Directive would partially address the gaping hole in the 2019 Proclamation, which overlooks the role of social media discourse in shaping the course of elections. There is still, however, an issue remaining with the narrow conception of the problem associated with social media platforms.

As highlighted above, dangers to the credibility and integrity of electoral processes come from different corners in the digital age. Social media's use of practices of registered political parties and candidates – whose conduct is and could be monitored by the NEBE – is unlikely to pose the greatest danger to the credibility of elections. Global actors, be it state or non-state actors, with geo-political or economic interests in the nation could marshal social media platforms to skew the elections. As was the case in other jurisdictions, sophisticated disinformation campaigns, including using sponsored political advertisements, pose serious dangers to the integrity of elections. Besides such foreign actors, Ethiopian political actors other than registered political parties could also use online platforms to influence the outcome. With the significant role of the Ethiopian diaspora in domestic political dynamics, social media platforms would simply offer new ways to reshape the domestic political reality, including election outcomes. The NEBE's regulatory responses thus far do not seem to have such new and sophisticated election-manoeuvring practices in mind. More crucially, no measure seems to be in the offing that would outline the duty or role of major social media platforms, such as Facebook, during the election cycle in Ethiopia.

Criminal law

Criminal law may also apply in some contexts. Title V (Crimes against Public Elections and Voting) of the Ethiopian Criminal Code touches upon an aspect of election disinformation.Footnote ⁷⁹ Article 471 criminalizes “unfair practices”, which include two key offences. One is campaign finance violations, for example collecting money or receiving donations contrary to campaign finance rules. The other offence concerns fraudulently securing, abusing or misappropriating votes directly or through an intermediary with (a) the help of false news or rumours circulated through, inter alia, the press, radio or “any media of publicity”; (b) the help of any other fraudulent, irregular or dishonest practices. This provision was drafted at a time when the spread of either election-related information or political adverts had not reached its present level of sophistication, but it specifically uses terms like “false news and rumour”, thereby capturing the essence of “disinformation”, and its illustrative list of media engaging in unfair electoral practices includes the new media, including social media platforms.

It is vital to note that this provision applies to anyone who engages in those practices and, as such, it is not restricted to political parties and candidates. In comparison, therefore, Ethiopian criminal law contains rules on election-related disinformation that are more progressive. These rules, however, do not provide adequate safeguards; for one thing, they offer remedies after the fact, and once the accused has engaged in the impugned acts, and hence do not preemptively prevent or mitigate the risk of disinformation. In addition, the transnational nature of the problem and weak institutional capacity to wade through the complexity of digital disinformation further diminishes the criminal law solution to the disinformation problem in Ethiopia, including in the election context.

Data protection law

In regulating the collection, processing and sharing of personal data, data protection law plays a key role in controlling the use of personal data for political purposes. Data protection law would constrain processing of sensitive personal data such as one's political opinion and profiling of data subjects. It would impose a series of transparency obligations on data controllers and processors, such as to clearly state the type of data collected, for how long, and its purposes as well as to whom it may be shared. These duties are embedded in a series of data protection principles and rights. On top of this, this area of law would put in place an oversight mechanism by which compliance is monitored. As the controversy over the 2016 US presidential elections illustrated, exploitation of personal data from social media users was part of the disinformation campaign.Footnote ⁸⁰ In the wake of this debacle, the debate has centred on the role of data protection law in regulating the use of personal data for political campaigning purposes. While Ethiopia does not have data protection legislation, it has recently issued a Bill that provides useful safeguards against exploitation of personal data for political purposes.Footnote ⁸¹

This Bill addresses some of the key issues of data protection in the context of elections. First, unlike the laws of many countries, Ethiopia's draft legislation does not exempt processing of personal data for political purposes.Footnote ⁸² This means that any collection, processing or dissemination of personal data by anyone for political purposes must comply with data protection principles and rights as well as the attendant regulatory controls. Second, it also categorizes personal data concerning a data subject's political opinion as “sensitive personal data”.Footnote ⁸³ And this would mean that while processing of such data is prohibited in principle, it is strictly regulated when certain exceptions warrant processing. Third, the Bill also addresses profiling – i.e. automated processing of personal data – which is a key aspect of current social media political advertisement practices.Footnote ⁸⁴ Profiling is permitted in only exceptional circumstances, but is totally prohibited if the processing of the data involves sensitive personal data. Fourth, the Bill also defines personal data in a manner that captures the state of the art in the world of online political advertisements, where use is mainly made of pseudonymized and anonymized data. The definition of personal data includes (pseudonymized or anonymized) data that could allow the identification of a person when read/combined with other data. Fifth, the draft law imposes a series of transparency obligations on data controllers and processors under the principles of “fairness and transparency”, “purpose limitation”, “data minimization”, and data rights such as the “right to be informed” and the “right of access”.Footnote ⁸⁵ Sixth, to monitor protection of data privacy, the Bill establishes a Data Protection Commissioner and also mandates the appointment of Data Protection Officers by data controllers. But the virtues of this impending law are constricted; for one thing, it is a mere draft and hence it is not clear when it will be enacted and will enter the statute book, however even after entry into force, its implementation will be impaired by a number of factors, including lack of resources and expertise. As the experience in other African countries suggests, it would be a while before a national data protection authority is up and running, which might mean after several elections and referenda have already been held and potentially marred by disinformation.

Advertisement law

Ethiopia has an advertisement law that regulates all forms of advertising in Ethiopia, including web advertising. Its conception of advertisement is, however, problematic. The definition of advertisement under the law suggests that it concerns only “commercial ads”, implying that adverts made for non-profit purposes are not covered,Footnote ⁸⁶ but again, an overall reading of the law suggests that it applies to all forms of adverts, including non-commercial adverts such as personal and political adverts. Adverts “having political goals” are listed among prohibited or restricted types of adverts, however the law carves out an exception for political adverts disseminated during the election season by political parties and individual candidates registered under Ethiopian law.Footnote ⁸⁷ This means that other adverts with certain political goals, but disseminated by actors other than those two groups, are prohibited. Theoretically, this would include sponsored adverts by foreign state and non-state actors, but in light of the law's overall narrow scope in regard to social media-based sponsored adverts, the law offers little in addressing adverse political advertising from overseas. It is also vital to note that the duty of monitoring the implementation of this legislation falls on the EBA. However, the Authority's regulatory remit thus far has been restricted to the regulation of legacy media – i.e. broadcast and print media, and this means that types of political adverts disseminated through social media platforms as tools of disinformation campaigns would not be addressed by Ethiopia's advertisement legislation.