Category: AI (artificial intelligence)

Does Africa Need Its Own Digital Markets Act? Key Takeaways from the Centre for Competition Law and Economics’ Webinar on Digital Competition Policy Developments in Africa

Posted on by Editor

By Michael-James Currie and Matthew Freer

On 9 June 2026, the Centre for Competition Law and Economics (“CCLE”) at Stellenbosch University convened a practice webinar that captured, in real time, the tensions, ambitions, and practical fault lines shaping digital competition policy across the African continent. The timing was deliberate. Across Africa, competition authorities have moved past the abstract question of whether digital markets require special attention. Instead, they are now wrestling with a harder set of questions: how to regulate, who should regulate, and, most exactly, what exactly the objectives of that regulation ought to be.

The webinar brought together three voices, each occupying a distinct vantage point. Professor Willem Boshoff, Co-Director of the CCLE, Department of Economics, Stellenbosch University. He opened with a survey of national and regional developments, sketching a landscape marked by innovation but also fragmentation. Malick Diallo, Head of Competition at the African Continental Free Trade Area (“AfCFTA”) Secretariat, then offered a rare first-hand account of how the continental body is positioning itself within that landscape. And finally, Michael-James Currie, Director at Primerio, brought the practitioner’s lens: what do these proliferating rules mean for clients trying to comply, invest, and compete in an environment where regulatory priorities remain dangerously unclear?

The South African Starting Point

Boshoff began by anchoring the discussion in the South African experience, not because it is representative of the continent, he was careful to say it is not, but because it offers a useful baseline for comparison. What is striking about the South African approach, he observed, is how the competition authorities have relied on existing tools rather than demanding a separate, bespoke digital regulatory regime. They have repurposed market inquiry tools, adapted merger control frameworks to capture killer acquisitions, and sought to develop broader skills across the authority rather than building a dedicated digital unit.

That last point is revealing. Boshoff noted, that running a competition authority in Africa comes with limited resources and scarce specialised skills. Building a standalone digital unit is expensive. Instead, the South African authorities have attempted to mainstream digital expertise across the organisation, relying on the two major market inquiries, the Online Intermediation Platforms Market Inquiry and the Digital Media Platforms Market Inquiry, to build institutional understanding from the ground up.

What is equally notable, Boshoff argued, is the preference for time-bound remedies and the distinctly developmental focus that runs through South African competition enforcement. Supporting smaller players, protecting local businesses, and ensuring that digital markets serve broader industrial policy goals have become central features of the approach. “In a sense,” he said, “this is quite different from the approach taken in the European Union, even though it might mean we do a bit more ex ante work within the competition law framework.” The EU has gone for strong, explicit ex ante regulation. South Africa has not, at least not yet.

But Boshoff was careful to emphasise that South Africa is not the continent. When you move beyond its borders, the picture changes dramatically.

Kenya, COMESA, and the March Toward Ex Ante Rules

Kenya represents a different trajectory. Boshoff described a jurisdiction that has historically taken a lighter-touch approach, not unlike South Africa’s. But recent developments, specifically the country’s e-commerce policy and the accompanying amendments to its competition legislation, signal a deliberate shift toward ex ante measures. The competition amendment bill includes alternative thresholds for digital mergers and rethinks how dominance should be assessed in a digital setting. Whether that will translate into dramatically different enforcement outcomes remains to be seen, Boshoff cautioned, but the fact that these provisions are being baked into the legislation itself is significant.

Move up one layer further, to the regional level, and the picture shifts again. COMESA, Boshoff noted, has been remarkably active. Its 2025 regulations align closely with the European DMA-style approach, complete with specific prohibitions, digital merger thresholds, and a posture toward prohibited practices that is far more prescriptive than South Africa’s case-by-case method. That comes with its own set of challenges, Boshoff acknowledged, both for enforcers and for the parties subject to those rules.

Across all these jurisdictions, however, Boshoff identified two common threads. The first is a merger of competition policy and consumer policy, not new, but particularly pronounced in the digital context, where exploitative conduct targeting specific groups of customers has become a focus of attention. The second is an emphasis on protecting small local players, whether through merger remedies or abuse of dominance enforcement. That emphasis on contestability, Boshoff suggested, raises a deeper question: is the goal to have two or three players competing head-to-head, or is it to build ecosystems where one or two large players create opportunities for many smaller ones in adjacent markets? Those are, in effect, industrial policy decisions baked into competition law. And they have not yet been fully debated.

The AfCFTA’s Role

If the national and regional picture is one of fragmentation and divergence, Malick Diallo’s contribution was an attempt to map how the AfCFTA intends to impose order without overriding legitimate local and regional autonomy. Diallo was clear from the outset: the AfCFTA protocol on competition policy was never designed to replace or supersede national or regional frameworks. The preamble explicitly recognises the central role that national and regional authorities will continue to play in promoting fair competition and inclusive growth in intra-African trade.

In describing what the continental body is for, Diallo explained the three-layer architecture. National authorities handle matters of a domestic nature, classic enforcement, abuse of dominance, local measures. Regional bodies like COMESA, ECOWAS, WAIMU, and SAMRC address cross-border conduct within their respective markets. And the AfCFTA Competition Authority steps in only where there is a “continental dimension”, defined in Article 1 of the protocol as conduct, practices, mergers, or agreements that have a significant effect on the markets of at least two state parties that do not share the same regional economic community jurisdiction.

Digital markets are the clearest illustration of why this matters. Diallo pointed to a study by the African Competition Forum showing that Google holds an estimated 90% market share in search across the continent. That dominance is felt in every African country simultaneously. A national authority can deal with purely domestic conduct, and a regional body can handle matters limited to its region, but when conduct cuts across different regions, or when no regional body has jurisdiction, the AfCFTA fills the gap. “We are filling the enforcement gap that arises in cross-regional and truly continental transactions,” Diallo said.

He identified five concrete ways the AfCFTA complements existing work:  

  1. It fills the jurisdictional gap.
  2. It promotes harmonisation of laws and standards. Diallo noted that COMESA has already adopted new provisions on abuse of economic dependence, aligning with the AfCFTA protocol, and the secretariat is supporting other state parties to do the same.
  3. It has established the AfCFTA Competition Network (AFCNet), a platform for regular dialogue, case referrals, joint investigations, and the development of common approaches to market definition, data access, and remedies.
  4. It facilitates capacity building, allowing more advanced jurisdictions like South Africa to share expertise with younger ones.
  5. It provides a structured channel for information sharing, including confidential information, to avoid the inconsistencies and duplicative interventions that currently plague the system.

“What we are trying to build is not a parallel enforcement regime,” Diallo emphasised, “but a continent-wide ecosystem, one where national authorities handle domestic cases, regional bodies handle cross-border intra-regional cases, and the AfCFTA handles truly continental conducts.” Digital markets, given their cross-border nature, network effects, and tendency toward gatekeeper dominance, are the clearest illustration of what the continental authority is meant to tackle.

The Practitioner’s Warning

Speaking from the perspective of a competition lawyer advising clients who must navigate this proliferating regulatory landscape, Michael-James Currie raised a series of pointed questions about whether the current wave of rulemaking is outpacing the capacity of authorities to enforce those rules wisely.

He began with killer acquisitions. Many jurisdictions have lowered their merger thresholds to capture these transactions. But Currie asked a deceptively simple question: what happens after the transaction is captured? Are agencies actually able to make informed, forward-looking assessments in dynamic markets? He noted that while the theories of harm in killer acquisitions are well established, it would be illuminating to conduct an ex post assessment of all the digital mergers captured by lower thresholds over the past ten years. How many of them, with the benefit of hindsight, ought to have been prohibited? The Facebook-Instagram decision is often cited as a cautionary tale, Currie acknowledged, but even there, one must ask: would Instagram be where it is today without Facebook’s investment and synergies?

That question is not merely academic. It goes to the heart of whether new rules are solving a real problem or simply increasing regulatory friction. It leads directly to the issue of capacity. Even the most resourced jurisdictions struggle to make accurate forward-looking assessments in digital markets, Currie observed. For African authorities, many of which lack dedicated digital units, the challenge is magnified. “It’s very difficult,” he said, “not even for a jurisdiction that lacks the resources, even for the most resourced jurisdictions and agencies who have been looking at digital markets for many years, it’s tough for them too.”

Currie then turned to the issue of gatekeepers, which he described as “just one element of digital markets.” He noted that COMESA is currently drafting regulations to define what a gatekeeper is, a process that will not be uncontentious. He pointed to a recent decision where Meta successfully overturned a European Commission designation of Facebook Marketplace as a gatekeeper under the Digital Markets Act, with the General Court of the European Union ruling in Meta’s favour. That decision, Currie argued, shows that there will be a great deal of litigation over who qualifies as a gatekeeper, and that authorities will have a very tough time defining the relevant product markets in which a respondent is said to be a gatekeeper.

Perhaps most provocatively, Currie suggested that the policy conversation is disproportionately focused on platforms and gatekeepers while neglecting digital infrastructure. Currie suggests that if one wants to grow local industries, digital infrastructure is critical. The attention given to platforms, he argued, comes at the expense of the underlying infrastructure that would enable local players to compete in the first place.

It was in the context of competing policy objectives, however, that Currie delivered his most pointed remarks. He observed that South Africa has always mixed industrial policy into its competition regime, protecting employees, supporting SMEs, promoting historically disadvantaged persons. “It all sounds very good on paper,” he said. “But it is very difficult for an agency or regulator, or even government, to say, if there’s a tension between what’s good for consumers and what’s good for a certain class of competitors, who will we prioritise?”

That question is not abstract. It arises in real cases, and it requires an answer. Currie’s concern was that regulators have not provided one. Instead, they have effectively said: trust us. We will arrive at the right conclusion. We don’t want to harm innovation or investment. Just trust us.

“That is a very difficult message to sell to industries and stakeholders,” Currie said. “Policymakers and regulators need to set out, very clearly and deliberately, what they prioritise over what under instances of tension.”

The Unresolved Question

Boshoff, returning to the discussion, noted that Currie’s concerns connected directly to a deeper issue that the webinar had only begun to explore. The implicit industrial policy focus of digital market regulations across the continent has not yet grappled with how best to support African platforms and ecosystems. The EU policy debate, Boshoff observed, is currently centred on digital mergers, scaling, and how merger policy might support European-based platforms in response to the Draghi report. That debate is largely absent in Africa.

Conclusion

The webinar left little doubt that Africa is moving rapidly toward a multi-layered digital competition regime, with the AfCFTA positioning itself as the essential capstone. Malik Diallo’s contribution was valuable precisely because it came from inside the process, he was able to articulate not only the legal architecture but the practical mechanisms, AFCNet, harmonisation efforts, capacity building, through which the AfCFTA intends to make that architecture work.

Currie’s warnings were however valuable. Regulation without clarity of objective is not sound policy. Asking stakeholders to trust that regulators will balance consumer welfare, SME protection, industrial development, and innovation in every case is not a sustainable basis for compliance or investment. As African authorities continue to build out their digital competition frameworks, whether at the national, regional, or continental level, they would do well to answer the question Currie posed. When tension arises between competing objectives, what comes first?

Until that question is answered clearly and deliberately, the risk is not that African competition policy will be too strong or too weak. It is that it will be unpredictable. And for businesses trying to invest and compete across the continent, unpredictability is its own kind of harm. However, the message is equally not one of despair but of opportunity: African competition authorities are building something unprecedented, a truly continental enforcement dialogue, and if they can answer the hard questions about what they value most, they may yet produce a model for digital regulation that is as dynamic as the markets it seeks to govern.

‘Antitrust-Adjacent’: Kenya’s Artificial Intelligence Bill, 2026: Building a Practical and Coherent Framework for AI Regulation and Governance

Posted on by Editor

In this new series, entitled “Antitrust-Adjacent,” AAT covers relevant developments of interest & adjacent to competition-law issues. Our first installment on AI Regulation is co-authored by Kenya Primerio Partner Fidel Mwaki and Alfred Nyaga.

By Fidel Mwaki & Alfred Nyaga

Introduction

Artificial Intelligence (AI) is rapidly reshaping Kenya’s digital economy. From financial services and telecommunications to healthcare, logistics and digital platforms, AI systems increasingly underpin critical decision-making and service delivery across both the public and private sectors.

The introduction of Kenya’s Artificial Intelligence Bill, 2026 (the Bill) marks a significant milestone as the country seeks to articulate a comprehensive regulatory and governance framework for AI.

However, as Kenya moves toward formalising this framework, several foundational questions arise. Does the Bill effectively regulate AI systems? How should institutional oversight be structured? What would a practical regulatory model look like in the Kenyan context? And what governance architecture is required to ensure responsible, transparent and innovation-friendly deployment of AI?

Key Recommendations for Policymakers and Industry

To strengthen the effectiveness and practicality of Kenya’s Artificial Intelligence Bill, 2026, the following considerations may be useful:

  1. Establish clear and operational risk classification criteria


Define objective and measurable criteria for determining what constitutes a high-risk AI system. This should enable developers and deployers to assess compliance obligations at the design stage and reduce regulatory uncertainty.

  • Introduce a framework for general-purpose AI (GPAI) models


Recognise the growing role of GPAI models as foundational infrastructure and consider tailored obligations around transparency, accountability and safe deployment.

  • Adopt a sector-sensitive regulatory approach


Different sectors present different AI risks. The framework should enable coordination with sector regulators such as finance, telecommunications and healthcare to ensure context-specific oversight.

  • Clarify liability and accountability across the AI lifecycle


Establish a clear allocation of responsibility between developers, deployers and users of AI systems, particularly where systems are used in decision-making with real-world consequences.

  • Strengthen institutional coordination mechanisms


Provide clear guidance on how the proposed AI Commissioner will coordinate with existing regulators, including the ODPC, CA, CBK, CAK and other sector bodies, to avoid duplication and regulatory fragmentation.

  • Provide for independent oversight and audit mechanisms


Introduce provisions for AI audits, documentation standards and oversight processes, particularly for high-risk systems and sensitive applications.

  • Embed flexibility through phased and adaptive regulation


Allow for the framework to evolve through secondary regulations, guidelines and regulatory sandboxes, ensuring responsiveness to technological developments without creating uncertainty.

These measures would help ensure that Kenya’s AI regulatory framework is practical, coordinated and capable of supporting both innovation and responsible deployment.

Understanding the Regulatory Gaps: Does the Bill Effectively Regulate AI Systems?

The Bill places considerable emphasis on high-risk AI systems. While this approach mirrors developments in jurisdictions such as the European Union, it is not immediately clear how developers or deployers are expected to determine whether an AI system falls within that category.

In practice, this classification becomes one of the most important compliance questions for organisations building or deploying AI systems. Without clear and predictable criteria, organisations may struggle to assess their obligations. This uncertainty has the potential to affect both compliance and innovation.

Equally important is the treatment of AI systems that fall outside the high-risk category. AI technologies can generate meaningful societal and economic risks even where they are not formally classified as high risk. Issues such as misinformation, manipulation, algorithmic bias and systemic economic disruption may arise from such systems.

The Bill also appears to focus significantly on public sector deployment of AI systems, yet much of the development and deployment of AI in Kenya currently occurs in the private sector. Industries such as financial services, telecommunications, logistics and digital platforms already rely heavily on AI-driven systems. A balanced regulatory approach should therefore account for both domains.

A further gap is the absence of a clear framework addressing general-purpose AI (GPAI) models, including systems capable of generating text, code, images and other forms of content. These models increasingly serve as foundational infrastructure for a wide range of downstream applications and may require tailored regulatory treatment.

Globally, regulators are beginning to address these issues more directly. The EU AI Act introduces detailed risk classification frameworks and obligations for developers and deployers of high-risk systems, while also addressing GPAI models. China’s evolving regulatory framework similarly addresses algorithmic transparency, registration of AI deployers and security assessments prior to deployment.

The Governance Question: How Should AI Oversight Be Structured?

The Bill introduces an important institutional feature through the establishment of an AI Commissioner, tasked with overseeing AI development and deployment, monitoring compliance and issuing guidance on the responsible use of AI technologies in Kenya.

The creation of a dedicated oversight authority reflects an important recognition that AI presents regulatory challenges that extend beyond traditional legal frameworks.

However, the Bill raises a broader governance question: how should this oversight function interact with Kenya’s existing regulatory institutions?

Kenya already has several regulators whose mandates intersect with AI governance, including the Office of the Data Protection Commissioner (ODPC), the Ministry of ICT and Digital Economy, the Communications Authority of Kenya (CA), the Kenya Bureau of Standards (KEBS), the Competition Authority of Kenya (CAK), the Central Bank of Kenya (CBK), and various sector-specific regulators.

The Bill does not yet clearly define how the proposed AI Commissioner will coordinate with these institutions. In practice, effective governance of AI will require structured collaboration across multiple regulators rather than reliance on a single oversight authority.

There are also broader concerns around transparency and accountability across the AI lifecycle. AI systems typically involve multiple actors, including developers, deployers and organisations relying on system outputs. A coherent framework must therefore clearly allocate responsibility when AI systems produce harmful or unintended outcomes.

In addition, several aspects of operational governance are left to future regulations. These include independent AI audits, organisational governance frameworks, model transparency requirements and coordination mechanisms. While this approach provides flexibility, it also means that the governance architecture remains only partially defined at the legislative stage.

Designing a Practical Regulatory Framework for Kenya

These gaps point to a broader question: what would a practical and workable AI regulatory framework look like in the Kenyan context?

AI regulation globally is still evolving. Jurisdictions such as the European Union, China and Singapore are adopting different approaches, creating an opportunity for Kenya to design a framework that reflects both international best practice and local priorities.

A practical regulatory framework would benefit from several structural elements:

  1. A clear and predictable risk classification approach


Regulatory obligations should be tied to well-defined categories, enabling developers and deployers to assess their obligations at the design stage and reducing uncertainty.

  • Recognition of general-purpose AI models


These systems increasingly function as foundational infrastructure and may require tailored transparency, accountability and safety obligations.

  • Sector-sensitive regulation


Algorithmic risks vary significantly across industries, and systems deployed in sectors such as healthcare, financial services or critical infrastructure raise different regulatory concerns. Coordination with sector-specific regulators will therefore be necessary.

  • Clear allocation of responsibility across the AI lifecycle


AI systems often involve multiple actors, and a coherent framework must assign responsibility across this lifecycle to prevent regulatory gaps.

  • Building an Effective Governance Architecture for AI Oversight

Beyond identifying the governance gaps in the Bill, a key question is how Kenya can structure an effective and coordinated AI governance architecture in practice.

The introduction of an AI Commissioner is an important step toward institutionalising AI governance. However, effective oversight will require more than the creation of a single regulatory office.

AI intersects with multiple regulatory domains, including data protection, financial regulation, competition policy, communications regulation and consumer protection. Governance of AI systems will therefore require coordination across multiple institutions.

Organisational transparency and accountability will also be critical. As organisations increasingly rely on AI in decision-making, internal governance structures such as AI risk frameworks, audit mechanisms and oversight committees may become necessary.

At the same time, governance frameworks must ensure that regulation does not unintentionally discourage innovation. Mechanisms such as regulatory sandboxes and collaborative oversight models may help strike a balance between risk management and technological development.

Conclusion

Kenya now stands at a defining moment in shaping its AI regulatory and governance framework. The Artificial Intelligence Bill, 2026 provides an important starting point. However, its effectiveness will ultimately depend on whether the regulatory and governance framework is sufficiently clear, coordinated and capable of evolving alongside technological development.

The framework will need to reflect Kenya’s economic context and development priorities, while drawing from comparative approaches where relevant.

As the Bill moves forward, the upcoming public participation process presents an important opportunity for policymakers, technologists, legal practitioners and industry stakeholders to engage constructively with these issues and help shape a framework that supports both innovation and the responsible deployment of AI in Kenya.

  • Fidel Mwaki is the Managing Partner of FMC Advocates LLP (Kenya) and In-Country Partner (Kenya) at Primerio. He advises on corporate, regulatory and governance matters, with a focus on emerging issues in digital regulation and AI governance.
  • Alfred Nyaga is a Director at Digital Ethics Hub, a platform focused on shaping policy and practice in digital rights and AI regulation.