The AI We Cannot See: Why Africa Needs a Voice in the Rules Governing Artificial Intelligence

Artificial intelligence is being presented as the technology that will define the future. But for much of Africa, there is a more immediate question: who gets to decide what that future looks like?

Across the continent, people are already using AI to write, translate, study, create businesses, access information and make decisions. Governments are beginning to develop AI strategies. Companies are deploying automated systems. Young Africans are building products around technologies they did not create.

Yet the rules governing these systems are largely being shaped somewhere else.

That creates a problem that is bigger than technological inequality. It is a question of power.

If the data comes from Africa, the users are African, and the consequences are experienced by African communities, African voices cannot be an afterthought in the governance of artificial intelligence.

Africa is not waiting for AI

There is sometimes a tendency to talk about AI and Africa as though the technology has not arrived yet.

It has.

A university student in Lagos can use the same generative AI tools as a student in London. A small business owner in Nairobi can use AI to create marketing material. A developer in Accra can build an application on top of a model trained by one of the world’s largest technology companies. A doctor or health worker can increasingly encounter AI-assisted systems in professional settings.

The technology is already becoming part of everyday life.

The problem is that access to the technology does not automatically mean participation in the decisions surrounding it.

Africa can have millions of AI users without having a meaningful voice in how AI systems are designed, tested, governed and regulated.

That distinction matters.

Imagine a technology company develops an AI system that is highly accurate when interpreting English but performs poorly when dealing with an African language. Or an automated system used for employment, financial services or public administration produces worse outcomes for certain African populations. Who decides whether that level of performance is acceptable?

Who gets to define the problem?

Who gets to demand an explanation?

And, perhaps most importantly, who gets to change the system?

These are governance questions, not merely engineering questions.

The rules are becoming as important as the technology

For years, much of the conversation about Africa’s technological development focused on infrastructure.

Can people get online?

Is there reliable electricity?

Can businesses access affordable broadband?

Can developers obtain the computing resources they need?

Those questions remain important. But AI introduces another layer.

As artificial intelligence becomes involved in decisions that affect people’s jobs, finances, education, healthcare, security and access to public services, the rules surrounding AI become a form of infrastructure themselves.

A country can have fast internet and still have little control over the technologies operating through that infrastructure.

This is why AI governance matters.

Governance determines what companies can collect, how data can be used, what protections people have, how automated decisions can be challenged and who is responsible when something goes wrong.

And these decisions are increasingly being made through a combination of governments, technology companies, researchers and international institutions.

The danger for Africa is not necessarily that the continent will be excluded from AI.

It is that Africa could be included primarily as a market for technologies whose rules were designed elsewhere.

That would be a familiar pattern.

For decades, many African countries have consumed digital products developed in other parts of the world. The internet changed that dynamic somewhat by allowing African entrepreneurs to build businesses for global audiences. But AI introduces a new dependency because the most powerful systems are extraordinarily expensive to develop and operate.

A handful of companies control enormous amounts of computing power, data, infrastructure and technical expertise.

That concentration of power makes participation in AI governance even more important.

Data is not just information

One of the biggest reasons African participation matters is data.

AI systems depend heavily on data. The quality, diversity and representation of that data can influence how well systems perform for different populations.

But data is not simply an abstract technical resource.

It can describe people’s languages, identities, behaviour, preferences, locations, economic circumstances and interactions.

Africa therefore has something much more valuable than a collection of datasets. It has hundreds of millions of people whose lives are increasingly becoming part of the digital economy.

The question is: who benefits from that data?

If African data helps improve a commercial AI system developed elsewhere, African societies should have a voice in the conditions under which that data is collected and used.

This does not mean treating data as something that should never leave the continent. Nor does it mean preventing international research or technological cooperation.

It means recognising that data governance is connected to economic power.

A future in which African data continuously contributes to the development of global AI systems while most of the economic and technological value is captured elsewhere would reproduce an old problem in a new form.

The resource has changed.

The underlying imbalance has not.

And then there is language

Perhaps nowhere is the gap more obvious than language.

Africa is home to thousands of languages, yet many digital technologies have historically been built with a relatively small number of languages in mind.

AI has the potential to change this.

Machine translation, speech recognition and language models could make digital services much more accessible to people who have traditionally been poorly served by technology.

But that will not happen automatically.

If an AI model understands English exceptionally well but struggles with Hausa, Igbo, Yoruba, Amharic, Wolof or thousands of other African languages, the problem is not simply that the model needs a technical upgrade.

It tells us something about whose voices were present when the technology was being built.

Language determines who can interact with technology comfortably. It determines who can access information and who cannot. It can determine whether a person feels that a digital system was designed for them or merely made available to them.

AI governance therefore needs to include questions of linguistic representation.

Africa should not simply ask technology companies to make their existing systems slightly better at African languages.

African researchers and institutions should have a role in determining what gets built in the first place.

Representation cannot stop at conferences

There is another danger in the growing conversation about “African perspectives” on AI.

It can become symbolic.

African governments send representatives to international conferences. African researchers participate in panels. Technology companies launch initiatives on the continent. Reports are published about responsible AI in developing countries.

All of that can be useful.

But representation is not the same thing as power.

Having an African speaker on a panel does not necessarily mean African institutions influenced the decision being discussed.

True participation means having the ability to shape standards, challenge decisions, contribute research, negotiate rules and hold powerful organisations accountable.

Africa needs institutions capable of doing this consistently.

That includes governments, universities, technology companies, civil society organisations, journalists and independent researchers.

It also means investing in people who understand both sides of the problem: the technology itself and the social consequences of deploying it.

AI policy cannot be written entirely by people who do not understand how AI systems work.

But it also cannot be left entirely to engineers.

The most important questions are often not technical.

They are questions such as: Should this system be deployed? Who should have access to it? What happens when it fails? Who bears the cost?

Those are political and social questions.

African countries should not simply copy foreign AI laws

There is an understandable temptation for governments developing AI policies to look at what larger economies are doing and adopt similar approaches.

There is value in learning from other countries.

But copying regulation is not the same as building policy.

African countries have different economic structures, public institutions, infrastructure limitations and social realities.

A regulatory framework designed for a highly connected European economy may not address the same problems as one facing a country where millions of people remain offline.

Similarly, rules designed around large technology companies in the United States may not fully account for the realities of African startups, universities and public institutions operating with far fewer resources.

African AI governance should therefore learn from global developments without becoming subordinate to them.

The objective should not be to build an isolated African approach to AI.

It should be to ensure that African realities are part of the global conversation.

The opportunity is bigger than regulation

There is also a positive side to this conversation.

AI governance is not only about preventing harm.

It is an opportunity to decide what kind of AI ecosystem Africa wants to build.

African countries could use policy to encourage local research, open datasets, responsible innovation and African-language technologies.

Universities could become centres for AI research rather than merely consumers of AI products.

Governments could use public procurement to create opportunities for local technology companies.

Researchers could collaborate across borders to build datasets and models that reflect African populations.

Civil society could help ensure that communities affected by automated systems have mechanisms to question them.

And entrepreneurs could build AI products around problems that international companies may have little incentive to solve.

This is where governance becomes much more than regulation.

Good policy can create markets.

It can create research institutions.

It can create incentives.

It can determine who gets to participate in the next generation of technology.

The question of sovereignty

There is a bigger issue underneath all of this: technological sovereignty.

Sovereignty does not necessarily mean that every country must build its own giant AI model.

That would be unrealistic for many countries.

It means having enough capacity to understand, negotiate and influence the technologies on which society increasingly depends.

A country should not need to build its own search engine to have a say in how search affects its citizens.

Likewise, it does not necessarily need to build its own frontier AI model to establish rules around AI.

But it needs people who understand the technology.

It needs institutions capable of evaluating claims made by technology companies.

It needs researchers who can independently test systems.

It needs policymakers who can distinguish genuine innovation from marketing.

And it needs citizens who understand their rights when automated systems affect them.

Without that capacity, technological dependence can quietly become political dependence.

Africa must move from consumer to participant

The goal should not be to reject foreign AI.

That would make little sense.

African developers, businesses, researchers and ordinary users can benefit enormously from technologies developed elsewhere. Global collaboration is one of the great strengths of the digital age.

But collaboration works best when both sides have agency.

Africa should not have to choose between adopting AI and protecting its interests.

It should be able to do both.

That requires a shift in mindset.

Instead of asking only, “How can Africa benefit from AI?”, policymakers and technology leaders should also ask:

“What kind of AI should Africa help build?”

“What rules should govern it?”

“Whose interests should those rules protect?”

“Who gets a seat at the table when those rules are written?”

Those questions change the conversation from adoption to participation.

And participation is what Africa needs.

The future is being written now

Artificial intelligence is still young enough that many of its rules are being negotiated.

That window will not remain open forever.

Standards will be established. Business models will become entrenched. Governments will make decisions. Infrastructure will be built around particular technologies. Habits will form.

By the time the consequences become obvious, changing the underlying systems may be much harder.

Africa therefore cannot afford to wait until AI has completely transformed its economies and public institutions before becoming involved in its governance.

The continent needs to be present while the rules are being written.

Not as a guest.

Not simply as a market.

Not as a source of data.

But as a participant with its own interests, expertise and ideas about what responsible technology should look like.

The central challenge of the AI era may not be whether Africa can use artificial intelligence.

It almost certainly will.

The deeper question is whether Africa will have enough influence to shape the intelligence that increasingly shapes its future.

That is the conversation Africa needs to enter now.