Angola cut its internet for the first time in its recorded history on a Saturday in late July 2025.
The trigger was not an election or a war but street protests over fuel prices.
The authorities’ own figures put the death toll at 22; civil society monitors counted 30 dead and more than 1,500 arrests.
That episode sits inside a much longer pattern documented by Ephraim Percy Kenyanito, an international lawyer who has published a monthly record of technology, media and telecommunications law across Africa since January 2014.
In a report for Global Voices, he writes that the roundup reached its 152nd consecutive edition in August 2026.
Counting only entries where a government itself restricted, throttled or cut a network, and setting aside closures of individual broadcasters, he found more than 70 separate episodes across 32 countries, every year since 2014.
Access Now’s KeepItOn coalition counted independently and arrived at the same number, he writes.
The usual definition is inadequate
The debate over digital protest usually starts from hashtags, coordinated posting, online petitions and profile pictures changed in solidarity.
Kenyanito lists some of the continent’s most consequential campaigns: #ThisFlag and #ShutDownZimbabwe in 2016, #BringBackOurInternet during the Cameroon disconnection, #DataMustFall in South Africa, and #EndSARS in Nigeria in 2020 and #EndBadGovernance in 2024.
He argues that governments respond to online protest through two distinct instruments. The first targets a named individual: the state prosecutes the person who posted.
Kenyanito writes that he has found roughly 70 examples of that every year since 2014, citing Ethiopia’s Zone 9 bloggers in 2014, the co-founder of Jamii Forums under Tanzania’s Cybercrimes Act in 2016, Ignace Sossou in Benin in 2019, and two more journalists in Mali in June 2026.
Cybercrime statutes, false-news provisions and computer misuse acts are the tools, he writes, and the target is always a named individual.
The second approach is different. When many people gather, Kenyanito writes, the state prosecutes nobody and cuts the network instead.
Algeria, every June during exams
Not every shutdown is triggered by dissent. Algeria has cut the national internet every year during the baccalaureate exams to stop candidates circulating exam papers, and Kenya blocked Telegram during its own national examinations in 2023 and 2024.
Kenyanito notes that no crowd, protest or suppressed opinion was involved in those cases, and that an argument treating assemblies of dissent as the only trigger overstates its case.
But he draws a conclusion from the exam shutdowns: nobody claims Algeria switched off a national network to silence a view. The state reached for the network because the network was what people would use to coordinate.
A shutdown, in his framing, is not a speech instrument but an instrument for preventing coordinated action, and examinations and demonstrations are two instances of the same reach.
Cut the network during a protest, he writes, and you have not silenced a conversation. You have dispersed an assembly before it forms.
Article 9 or Article 11 of the African Charter
That reframing changes which provision of the African Charter on Human and Peoples’ Rights applies, Kenyanito argues.
Shutdowns are almost always litigated as freedom of expression problems under Article 9, while Article 11 protects the right to assemble freely with others.
The two clauses differ: Article 9(2) gives everyone the right to express and disseminate opinions “within the law,” whereas Article 11 permits “only necessary restrictions provided for by law.”
Necessity appears on the face of Article 11 and is absent from Article 9.
He notes that this is not an African peculiarity.
The International Covenant on Civil and Political Rights protects peaceful assembly in Article 21, and in 2020 the UN Human Rights Committee, which interprets the Covenant, issued General Comment 37 setting out what that article means.
Kenyanito writes that it reads Article 21 as covering assemblies held online and the organising of assemblies held in the street, and that on shutdowns it is explicit: states must not “block or hinder Internet connectivity in relation to peaceful assemblies,” nor engage in “geo-targeted or technology-specific interference with connectivity.”
He describes that as the settled reading for six years.
The African Commission’s 2017 Guidelines on Freedom of Association and Assembly go further, he writes, treating assembly as something a state must positively enable rather than merely refrain from crushing.
Its 2019 Declaration of Principles says at Principle 37(1) that states “shall facilitate” expression and access to information online. Principle 38(1) permits interference with content where it is “justifiable and compatible with international human rights law and standards.”
Principle 38(2), which prohibits “any disruption of access to the Internet … for segments of the public or an entire population,” carries no such qualifier.
Kenyanito calls the asymmetry deliberate and says it is doing far less work than it should.
He cites two examples of prolonged disruption.
Cameroon’s Anglophone regions lost access for more than a year from January 2017 in a bid to clamp down on dissent, and Fizi territory in South Kivu, Democratic Republic of the Congo, was offline for over a month from July 2026.
The Declaration is soft law, he acknowledges, which is a real limitation, but he calls it the clearest statement of the continent’s standard on access rights.
Litigators are not alone in leaning on Article 9, he writes. In 2024 the African Commission itself adopted a resolution on internet shutdowns during elections.
According to Kenyanito, the resolution treats a shutdown as harmful to expression and access to information under Article 9 and does not mention assembly at all.
The claim nobody argued
The clearest test of the assembly argument so far came from Senegal.
Ndiaga Gueye, who runs a Senegalese technology users’ association, took the government to the Economic Community of West African States (ECOWAS) Court of Justice after it cut mobile internet during the protests of June and July 2023.
He pleaded for expression and access to information under Article 9, the right to work under Article 15, and assembly under Article 11 of the African Charter and Article 21 of the Covenant.
On May 14, 2025, the Court ruled in his favour on grounds of expression, applying a four-part test of legality, legitimate aim, necessity and proportionality and holding that Senegal failed on all four: no law authorised the cuts, no evidence of the threat was produced, fewer restrictive measures existed, and a blanket shutdown was neither necessary nor proportionate.
It also ruled that the shutdowns violated his right to work.
His assembly claim went nowhere. Two applicants had brought it forward, the association and Gueye himself.
The Court dismissed the association’s assembly claim because it was not argued or supported.
It accepted that Gueye was entitled to bring his claim, but then decided the case on expression and the right to work and never ruled on whether the government’s shutdown violated his right to assembly.
General Comment 37 had been in force for five years by then.
Kenyanito’s reading of that outcome is that the assembly claim has been put to a regional court and gone nowhere, not because assembly is the wrong right, but because nobody argued it.
The judgments are thinner than they look
Court wins against shutdowns have so far delivered less than advocacy suggests, he writes.
Zimbabwe’s High Court set aside the January 2019 shutdown directives within days, but on the ground that the minister of state had no power to issue them, the president having reserved the administration of the Interception of Communications Act to himself; the court expressly declined the constitutional question.
The ECOWAS Court held Togo’s 2017 shutdown unlawful in June 2020, but because Togo conceded that no law authorised it, the Court never reached whether a shutdown could ever be necessary or proportionate.
In Kenya, Justice Bahati Mwamuye issued a conservatory order on May 14, 2025, restraining a shutdown pending hearing, and by June rights groups were demanding compliance with it.
Zimbabwe and Togo, in his assessment, are wins on legality rather than on the content of the right; neither tells a government what it may do once it has passed a law.
Senegal is the exception: he describes it as the first African ruling to hold that a shutdown fails a necessity and proportionality test, reached under Article 9 and without mention of Article 11.
Meanwhile the shutdowns continued. Tanzania cut the internet during its October 2025 election and Uganda ordered a blackout before the January 2026 vote.
Angola’s first shutdown in its history came 10 weeks after two courts ruled on the same day.
How to change the pattern?
Kenyanito’s prescription follows from the Senegal case. If litigators and human rights activists want to change the pattern, he writes, they must argue the assembly claim rather than merely list it.
Gueye’s case shows what happens otherwise: the right was named in the pleadings and disposed of for want of evidence and argument, while the court went to work on the claims that had been built.
The positive duty in the 2017 Guidelines, he argues, is harder for a government to argue around than a proportionality test on content, but only if someone puts it in front of a judge properly.
He also calls for naming the operators. Shutdowns are executed by licensed companies on instructions that are frequently unwritten and sometimes unlawful under the licensee’s own domestic law, he writes.
Civil society wrote to MTN, a leading telecommunications provider, about Sudan in 2019. License conditions, in his view, remain a lever barely pulled.
Finally, he argues that non-compliance should be treated as an enforcement failure rather than an open legal question. When a Kenyan court restrains a shutdown and the disruption continues, he writes, the deficiency is not in the law.
His conclusion is that the right to digital protest in Africa does not need inventing. It needs locating, and governments have spent 12 years showing where it sits: every time a crowd has gathered and the network has gone quiet.
Source: Global Voices
“,
“excerpt”: “An international lawyer’s monthly record of African technology and telecoms law counts more than 70 government-ordered network restrictions across 32 countries since 2014. He argues shutdowns are better understood as assembly bans than speech restrictions, and that courts have largely avoided ruling on that basis.”
}
Comments