Dr Blessing Vava
The Thursday Briefing. Adapted from my Bhawa Lectures talk at The Dome, Moto Republik, Harare, on 29 October 2026.
Open TikTok, Facebook or Instagram and look at the first post on your screen. You didn't ask for it. An algorithm had already picked it for you, based largely on what you watched yesterday, what you liked, what you skipped and how long you lingered.
Most of us rarely stop to think about this. We simply scroll, react and move on. Yet the same systems that choose our videos increasingly help decide which political messages reach voters, which stories dominate public debate and whose voices gradually drop out of sight.
During Zambia's election in August, I watched a fake GEN-20 results form spread across social media. I could clearly see people forwarding it, but I had no way of telling how much, if at all, the platforms' recommendation systems amplified it. We can watch misinformation circulate without ever knowing what the algorithms were doing behind the scenes, and that blind spot is part of the problem.
The same question arose during Brazil's October presidential election, when a dispute between electoral authorities and Meta prompted the electoral court to require equal treatment of the candidates' content. The dispute raised important questions about platform accountability, without proving deliberate algorithmic discrimination.
Across Africa, we have spent years celebrating cheaper smartphones and wider internet access. Connectivity, however, has moved much faster than public oversight. We can get online more easily than ever, yet the rules that decide what reaches us remain largely beyond our control.
We may have access to the platforms, but how much influence do we have over the rules that govern them?
Two terms are worth pinning down first. An algorithm is a set of instructions for completing a task, much like a recipe; for example, sorting names from A to Z. Artificial intelligence refers to computer systems capable of tasks such as recognising patterns, making predictions and generating responses. Many modern AI systems learn patterns from data rather than relying on a separate hand-written rule for every situation. Nobody had to write a rule telling TikTok to show Tendai a particular video. Its recommendation system draws on patterns in what people watch, skip and share. People still decide what the system should aim for, and that is where the question of control begins.
What digital sovereignty means
Digital sovereignty simply means having a real say over the technologies we depend on. It doesn't mean building everything ourselves, and it certainly doesn't mean handing governments more power over citizens. Take a foreign company that runs our national ID system. Sovereignty would mean being able to see how it handles our records, switch suppliers without losing them and quickly correct a record when it is wrong, while citizens remain free to challenge how the system is used. After all, a state can fully control its infrastructure while its people have almost no control over the state, and that is not the sovereignty I am arguing for.
Who decides what we see
I call one part of this the sovereignty of attention. Far more appears online every minute than anyone could possibly read or watch, so something has to decide what gets seen. Newspaper editors and radio producers once made many of those choices, and we at least knew who they were and could argue with them. Today, by contrast, invisible ranking systems make those decisions on a vastly larger scale.
As a result, silencing someone no longer requires banning a newspaper. A platform can simply leave a post online while showing it to fifty people instead of fifty thousand. The speaker remains free to post, but almost nobody hears them.
The real question, then, is who controls that visibility, and whether users or the wider public can challenge it. As I wrote in an earlier Thursday Briefing, 'The algorithm does not know whether you are looking for truth. It only knows that you are looking.'
What appears repeatedly in our feeds can also shape what we come to accept as normal. Schools have a hidden curriculum, the lessons children absorb beyond the timetable, and online feeds teach one too, through what they show and what they leave out. Some people keep appearing as experts while others are constantly presented as problems, and over time those patterns can turn somebody else's assumptions into common sense. Drawing on Antonio Gramsci’s concept of cultural hegemony, I use “mental hegemony” here to describe how algorithmic feeds can make particular assumptions and worldviews seem like common sense. As the Kenyan scholar Nanjala Nyabola reminds us, opening a conversation online does not by itself remove the inequalities that shape who gets heard.
Getting the platforms to account for their decisions is harder still. In October, X suspended more than 13,000 accounts over suspected foreign interference ahead of the US midterms. In Brazil, Lula's campaign alleged that Meta had restricted its reach, while Meta denied unequal treatment and said technical problems were responsible. The electoral court intervened, but that did not establish deliberate bias. Independent scrutiny remains difficult while much of the evidence sits inside the platforms.
That said, we should not respond by handing governments unchecked power over speech. Officials, too, must explain takedowns and allow people to challenge them. Otherwise, rules intended to counter falsehoods can easily become tools against criticism.
Why we end up with the technology we have
My doctoral research on Chinese ICTs in Zimbabwe led me to develop a framework I call geosociotechnopolitics. Debates about technological dependence often begin and end with foreign suppliers. Their influence is undeniably real, but on its own it does not explain why one system takes hold while another does not. To understand that, we also need to examine local needs, the design of the technology and the political decisions that enable its adoption. I bring those forces together through four questions:
• Geo: who is selling it, and what do they want?
• Socio: who is buying it, and what do they need?
• Techno: what does it actually do?
• Politics: who opened the door?
Consider the phone in your pocket. Zimbabweans once dismissed many Chinese phones as zhing zhong, meaning cheap and inferior. Yet in 2023, Transsion, which makes Tecno, Infinix and itel, accounted for roughly half of Africa's smartphone shipments. The company wanted new markets, and millions of Africans needed affordable phones. Features such as dual SIM slots, long battery life and cameras designed to capture darker skin tones met that demand well, and governments, crucially, allowed these products into their markets.
Neither low prices nor Chinese influence alone explains Transsion's success. These phones addressed everyday needs and reached customers through markets that governments had opened to them. That is what geosociotechnopolitics adds. It treats Africans as buyers, users and political actors who make their own choices, and my research on Chinese ICTs in Zimbabwe likewise found a constant interplay between foreign interests and domestic decisions.
Nick Couldry and Ulises A. Mejias call the wider pattern 'data colonialism', meaning data extraction that extends unequal power. My focus is narrower and more practical. I want to know how the four forces combine to make a particular dependency possible and, as a result, where choices can still be changed.
Sovereign in law, dependent in code
With electricity unreliable and data costly, some may reasonably ask why digital sovereignty should be a priority at all. Those are immediate barriers to getting online. Once a school, ministry or household comes to rely on a digital system, though, a different problem arises, and it concerns how much control we actually keep over it.
This is where geosociotechnopolitics meets what I call the governance divide. The framework explains how a system enters our lives, while the divide describes what happens once we depend on it and can use the system but cannot inspect it, repair it or renegotiate the contract. As I wrote in June, 'A country may remain sovereign in law while becoming dependent in code.' We can pass our own laws, yet crucial decisions about IDs, public records and elections may sit inside software controlled elsewhere. We own the house, but someone else holds the keys. Consequently, the time to negotiate access, accountability and the right to leave is before signing the contract, not once switching has become unaffordable.
Who thinks for Africa?
Dependence also affects where we turn for answers. Ask ChatGPT, Gemini or Meta AI on WhatsApp a question, and you will usually get a fluent answer within seconds. Fluency alone, of course, is no proof of accuracy. In 2023, two New York lawyers submitted court cases supplied by ChatGPT that looked entirely convincing but did not exist. The judge fined the lawyers and their firm US$5,000, since checking the sources was their responsibility.
Language brings the problem closer to home. Ask a translation tool to explain the Shona proverb Chara chimwe hachitswanyi inda, meaning one thumb alone cannot crush a louse, and the words may well appear while the meaning disappears. That is amusing with a proverb but genuinely dangerous with medical advice or a legal notice. Ngũgĩ wa Thiong'o argued that language carries culture, memory and identity, so when AI struggles with Shona, Ndebele or Tonga, we should naturally ask whose knowledge went into its training.
We need researchers who can test these systems in African languages and against African histories, especially before they are used in classrooms or public services. Fortunately, that work is already under way. Masakhane is a grassroots community whose name means 'we build together' in isiZulu, and South Africa's Lelapa AI has developed InkubaLM for isiZulu, isiXhosa, Swahili, Yoruba and Hausa. Shona, Ndebele and Tonga deserve the same attention.
The information we hand over
We often share sensitive information without thinking about where it goes. Someone pastes a confidential report into an AI assistant for a quick summary, a teacher enters pupils' records, and a professional uploads a client's details. Whether that information is retained or used to train a model depends on the service and its settings, and few users ever check before pressing send.
Political information carries particular risks. The Cambridge Analytica scandal exposed the harvesting of data from as many as 87 million Facebook users through a personality quiz for political profiling. The company was also active in African politics. Its executives discussed work in Kenyan elections, while reporting linked it to Nigeria's 2015 election. More recently, on 21 October 2026, in a New Mexico consumer-protection case connected to the Cambridge Analytica revelations, a New Mexico judge ordered Meta to pay US$25 billion after a jury found it had misled users. Meta says it will appeal.
Our national ID databases, clinic records and tax files deserve equally serious scrutiny, and we should demand answers from our own governments as firmly as we do from Meta or Google.
I was reminded of this at Robert Gabriel Mugabe International Airport, when I used the self-service passport scanner. It worked in seconds, but I later wondered where that information went and who could access it. The scanner may not use AI, but it is part of the digital machinery on which public services now depend. A foreign supplier is not evidence of wrongdoing. Even so, we have a right to know who can see our records and who can correct a mistake when a flight is boarding.
Who answers when it goes wrong
Kenya shows what happens when institutions decide to act. In 2023, Worldcoin, a project co-founded by OpenAI's Sam Altman, set up biometric-scanning 'orbs' in Nairobi, and thousands queued to have their eyes scanned in exchange for crypto tokens worth about US$50. Kenya suspended the operation that August. Then, in May 2025, the High Court ordered Worldcoin to delete the data it had collected from Kenyans, under the Data Protection Commissioner's supervision. The intervention ultimately depended on a law, a regulator willing to act, and a court able to enforce its decision.
New Mexico's action against Meta points the same way, showing that a jurisdiction can require a global platform to answer legal claims. Neither case proves that regulation always succeeds, but both show why credible regulators, courts and enforceable rights matter. Zimbabweans use the same platforms, so it is fair to ask whether our own institutions have the independence and resources to hold them to account.
Nor is African ownership a guarantee. A locally owned platform can suppress dissent or misuse personal data just as readily as a foreign one. Sovereignty should accordingly be judged by what people can know, contest and change, whoever happens to own the system.
Where Africa stands
Africa's choices are constrained in three ways, by infrastructure, expertise and enforcement. Take infrastructure first. Huawei has built significant parts of the continent's telecoms networks, while Meta and Google have invested in undersea cables. Such investment can certainly expand access, but it is worth asking how much say we retain over services that are difficult to replace.
The second constraint is expertise. Suppliers now come from China, America, Europe, India, the Gulf and Africa itself, but having options helps only if our universities, regulators, public officials and journalists can properly evaluate the technology and the contracts.
The third is enforcement. Zimbabwe has a Cyber and Data Protection Act, with POTRAZ serving as the data protection authority, but whether it can effectively investigate and challenge a global platform is still an open question. Across Africa, 39 countries now have data protection laws, and their value ultimately depends on whether people can use them to seek redress.
The free AI test
Suppose a company offered every public school in Zimbabwe a free AI assistant. Most of us would be tempted to say yes on the spot. It is worth slowing down first.
I would start by asking what the company gets out of it. It may genuinely want to help, but free products have a way of opening markets, collecting data and making the next contract hard to turn down, so it pays to know what comes after the gift.
Then I would think about the schools themselves. Not the ones in Harare where the launch would happen, but a school in Binga or Mutoko, where the electricity comes and goes, forty pupils share a single device and many children ask their questions in Tonga, Ndebele or Shona. A polished demonstration in the capital tells us very little about whether the tool will help in a classroom like that.
I would also want to know what the assistant actually does, whether children learn more with it, where it gets things wrong, and what they might stop learning to do for themselves if they lean on it.
Finally, I would ask who said yes, and on what terms. A ribbon-cutting ceremony cannot tell us whether anyone at the ministry read the contract carefully or looked at other options.
Those are the four questions at work, and engineers and officials cannot answer them alone. Teachers, parents and learners should have a say, because they are the ones who will live with the decision.
How we fight for digital sovereignty
Africa is hardly alone here. Even Europe, with its wealth and strict data rules, depends on external suppliers for much of its digital infrastructure and services, and the European Parliament has discussed estimates that Europe relies on non-EU providers for over 80 per cent of its digital products, services, infrastructure and intellectual property. In June 2026, the European Commission proposed measures to strengthen technological sovereignty, covering chips, cloud computing, AI and open-source alternatives, and Europe has set a target of a 20 per cent share of the global chip market by 2030. The European Parliament has even moved from Google towards the French search engine Qwant. We cannot simply copy Europe's approach, but its debate shows that control over technology is a political choice, and one that even wealthy regions have to fight for.
Merely criticising Silicon Valley and Beijing is not a plan, and companies will not surrender control of profitable systems simply because we ask. At the same time, replacing foreign influence with unchecked state power would solve little. Digital sovereignty must give African governments the capacity to negotiate with global technology companies while ensuring that citizens can question how those technologies are used.
We can begin with the contracts our governments sign. When a ministry purchases facial recognition technology, a digital ID system or an AI service, the public should know who supplied it, what it costs, who can access the information and what happens when the contract expires. We should also be asking whether the government can change suppliers without losing access to its own data. These are questions that ought to be settled before contracts are signed, not when a country discovers that leaving has become too expensive.
But transparency means little without people who understand the technology. Africa needs computer scientists, linguists, lawyers and researchers who can independently examine the systems being introduced into our societies. A Zimbabwean university could test an AI assistant by asking it 200 questions in Shona about farming, health, education and the law, then publish the answers and explain the mistakes. We would have evidence of how well it serves Shona speakers, beyond what its developers promise.
We also need to make better use of the laws and institutions we already have. South Africa's Protection of Personal Information Act and Kenya's Data Protection Act provide safeguards concerning certain forms of automated decision-making. Ahead of South Africa's 2024 elections, its Electoral Commission worked with major technology platforms to address disinformation. These arrangements matter, but their value ultimately depends on whether citizens can challenge decisions, regulators can investigate complaints and courts can enforce the law. Passing legislation is one thing. Making it work for ordinary people is another.
There is also a strong case for regional cooperation. Why should Zimbabwe, Zambia and Malawi negotiate expensive technology contracts separately when they face many of the same challenges? Through SADC and the African Union, countries could pool technical expertise, agree on minimum standards for public technology procurement and negotiate better terms with suppliers. The African Union's Continental AI Strategy already provides a foundation for cooperation on AI governance, infrastructure, research and skills development. We often say chara chimwe hachitswanyi inda. The proverb is just as relevant to digital governance as it is to everyday life.
None of this will amount to much if digital sovereignty simply transfers power from foreign corporations to a small group of political and economic elites at home. An African-owned platform can suppress dissent, and a government-controlled database can be used to monitor citizens. What matters is whether people have the right to know how these systems operate, question decisions made about them and seek redress when something goes wrong.
Ultimately, we need governments strong enough to negotiate with global technology companies, but also citizens and institutions strong enough to hold those governments accountable.
So, who owns the algorithm?
Owning data means little if people have no say in how it is collected, read and used. The real questions, therefore, are who benefits when a system makes a decision, who can challenge that decision, and who has the power to put things right when it goes wrong.
Next time you open your phone, look at the first video in your feed. Someone built the system that put it there and decided what it should reward. Those same choices, however, matter far more when software shapes what children learn in class, what a nurse sees on a hospital screen, or which political messages reach voters.
Decisions like these should not be left to technology companies, nor should they be handed to governments that answer to no one. Instead, we need the expertise to question the systems we buy, the bargaining power to set fair terms and institutions strong enough to enforce them. Above all, people need to be able to challenge the decisions that shape their lives. Indeed, if Africans cannot question the systems that govern their digital lives, whose sovereignty are we really defending?
Dr Blessing Vava is a Zimbabwean governance researcher. He holds a doctorate in Communication Studies from the University of Johannesburg, where his research examined the diffusion of Chinese information and communication technologies in Zimbabwe and their implications for politics and society. He developed the concept of geosociotechnopolitics to explain how geopolitics, technology, social conditions and domestic political choices shape technological adoption. He is the founder of the Southern Africa Coalition for Democracy and Accountability (SACDA), where his work focuses on democracy, electoral governance, digital sovereignty and regional accountability
Further reading and key references
Couldry, N., & Mejias, U. A. (2019). The Costs of Connection: How Data Is Colonizing Human Life and Appropriating It for Capitalism. Stanford University Press.
Cyber and Data Protection Act [Chapter 12:07] (Zimbabwe), 2021.
Data Protection Act No. 24 of 2019 (Kenya), section 35.
Lelapa AI (2024). InkubaLM: A small language model for low-resource African languages.
Masakhane.
Ngũgĩ wa Thiong'o (1986). Decolonising the Mind: The Politics of Language in African Literature. James Currey.
Nyabola, N. (2018). Digital Democracy, Analogue Politics: How the Internet Era Is Transforming Kenya. Zed Books.
Protection of Personal Information Act 4 of 2013 (South Africa), section 71.
Vava, B. (2024). All Weather Friend? Chinese ICTs in Zimbabwe as General Purpose Technologies. DLitt et Phil thesis, University of Johannesburg.
Vava, B. (2026, 24 June). Beyond Infrastructure: Digital Sovereignty and the Politics of Ownership in Africa's Technomultipolar Age.
Vava, B. (2026, 15 October). Elon Musk's X Suspends 13,000 Accounts, Raising Questions for African Elections.
Vava, B. (2026, September). When the Algorithm Decides What We See: Who Governs Digital Democracy?
Vava, B. The Passport, the Database and the State: Who Controls Digital Zimbabwe?
Vava, B. Who Thinks for Africa? AI, Geosociotechnopolitics and the Battle for Cognitive Sovereignty.
No comments:
Post a Comment