AI Governance Fragmentation Cause For Concern

Kenya is moving swiftly to build an artificial intelligence (AI) governance framework, with potentially significant implications for any business that develops, deploys or relies on AI in Kenya.

In a relatively short space of time, a national AI strategy, multiple draft policies, sector-specific regulatory guidance and a private member’s AI Bill have been introduced. The volume of activity speaks to ambition and momentum, but it also raises concerns about harmonisation.

These instruments have emerged from multiple institutional sources and, in some instances, propose different governance architectures for overlapping areas. Complicating matters, policy and legislation have so far moved in parallel, a departure from the norm.

Unless these instruments are deliberately harmonised as they progress, Kenya ‘s AI governance framework risks having overlapping mandates, duplicated obligations and legal uncertainty.

Laying The Groundwork Through Strategy

The Government set the ball rolling in March 2025 when the Ministry of Information, Communications and the Digital Economy launched the National Artificial Intelligence Strategy 2025-2030.

The strategy aims to provide a roadmap for responsible AI development and position Kenya as a leader in AI research, innovation and commercialisation, while aligning with the country’s broader digital economy agenda. It recognises that legal and regulatory development will be needed but places its emphasis on the foundations required for responsible AI adoption, including infrastructure, data readiness, skills development, investment, ethics, equity and inclusion.

Building on Kenya’s existing digital economy policy framework, the strategy was shaped through a consultative process drawing on input from the Government, development partners, the private sector, academia and civil society.

Rather than creating binding obligations, it favours a phased approach, with dedicated AI legislation envisaged as the AI ecosystem matures. That strategy-led approach has since been given institutional backing: in June 2026, Cabinet established a Standing Cabinet Committee on Artificial Intelligence to steer implementation of the National AI Strategy and coordinate AI policy across Government.

A Broader Policy Framework Takes Shape

Since the launch of the National AI Strategy, the policy framework has continued to take shape at pace, including two key cross-cutting draft policies emerging alongside sector-specific policies.

In July 2026, the Ministry of Information, Communications and the Digital Economy published the draft Kenya Artificial Intelligence and Other Emerging Technologies Policy, 2026 for public participation, which closed on 11 August 2026.

The draft policy goes beyond risk classification to address governance across the entire AI lifecycle, from design and deployment through to monitoring and accountability. A defining feature is its emphasis on ‘Sovereignty and Strategic Autonomy’, which it identifies as a standalone policy pillar aimed at strengthening domestic control over critical data, compute and AI infrastructure and reducing dependence on foreign-controlled infrastructure. This is reflected in proposed measures including data localisation for sensitive and strategic public-interest datasets and a sovereign cloud and computing strategy.

Two features of the draft AI Policy deserve particular attention from businesses.

First is its proposed broad extraterritorial scope, extending to foreign AI providers whose systems are procured, deployed or relied upon in Kenya, or whose outputs are used in or have direct and foreseeable effects within the country. As drafted, the policy sets no clear threshold based on whether a system is intentionally directed at the Kenyan market.

Second, the policy proposes a shared accountability model and indicates that future implementing legislation will allocate liability, oversight and redress across developers, deployers, operators, vendors and users, rather than resting responsibility with a single actor.

In parallel, the Ministry has advanced a draft National Data Governance Policy, 2026, on which public consultation closed on 5 June 2026.

It frames data as a national asset and extends the governance framework beyond personal data, which is already regulated under the Data Protection Act, 2019, to non-personal and aggregate data, signalling a broader approach to national data governance. The draft policy also proposes a National Data Governance and Emerging Technologies Council and a National Chief Data Officer.

The draft National Data Governance Policy shares the draft AI Policy’s emphasis on data and model sovereignty. It contemplates data localisation, a tiered sovereign-cloud strategy requiring domestic hosting for critical workloads, and greater domestic control over compute and infrastructure.

A further policy strand emerged in October 2026, with the State Department for Science, Research and Innovation publishing the Draft National AI Policy for Science, Research and Innovation for public participation, adding a more sector-specific policy track focused on AI in the science, research and innovation ecosystem.

Enter The AI Bill

Meanwhile the draft AI Bill was proposed earlier this year.  As drafted, the Bill would classify AI systems in terms of unacceptable risk, high risk, limited risk or minimal risk, with the most stringent governance, transparency, data-protection and record-keeping obligations proposed to apply to high-risk systems.

Beyond high-risk systems, the AI Bill would impose transparency and disclosure duties on all providers and deployers regardless of risk level, require the labelling of certain AI-generated synthetic media, criminalise the creation and distribution of harmful non-consensual uses of synthetic media, and mandate workforce impact assessments where AI systems are likely to affect employment.

The AI Bill would also establish an independent Office of the Artificial Intelligence Commissioner. This is one of several points at which the Bill overlaps with the draft AI Policy, which proposes a new National AI and Other Emerging Technologies Council to serve as the central coordinating and regulatory body. The overlap raises a question of institutional coordination as Kenya’s policy and legislative tracks continue to develop concurrently.

The parallel development of policy and legislation has since drawn closer scrutiny, with the Senate Standing Committee on Information, Communication and Technology recommending in October 2026 that the Senate should not proceed with the Bill until a comprehensive national AI policy is finalised. The Bill’s prospects are therefore uncertain. Although it formally remains before the Senate, it appears unlikely to proceed in its current form, and the immediate priority is now a national AI policy framework to guide any future legislation.

Sector-Specific AI Requirements

At the sectoral level, the Media Council’s Code of Conduct for Media Practice, 2025 requires media practitioners to disclose AI-assisted editorial material, keep AI-generated content under human review and refrain from cloning a real person’s likeness, voice or persona without consent.

The banking sector is also moving towards AI-specific guardrails. In September 2026, the Central Bank of Kenya published a Draft Guidance Note on AI in the Banking Sector, which would establish mandatory guardrails for AI and machine learning adoption by licensed financial institutions.

At the standards level, the Kenya Bureau of Standards’ Code of Practice for AI Applications (KS 3007) offers voluntary technical and governance guidance drawn from international and national frameworks.

On the copyright front, the draft Copyright and Related Rights Bill, a legislative proposal by the Kenya Copyright Board, would repeal and replace the Copyright Act, 2001. The Bill contains a narrow text and data-mining exception permitting copies of works to be made for computational analysis, although the exception is confined to non-commercial research and does not expressly address the use of copyright works to train commercial AI systems.

As the draft has not yet been formally tabled Parliament, its treatment of AI-related questions remains preliminary. It will also need to be considered alongside the broader Kenya Intellectual Property Bill, 2026, which is already before the National Assembly and would replace the existing copyright framework as part of wider reforms to Kenya’s intellectual property regime.

Developments In The Courts

The courts are becoming an important testing ground for AI governance.  The Judiciary’s draft Artificial Intelligence Policy proposes a risk-based approach to AI in court processes, including disclosure and verification of AI-assisted work. Recent cases show that courts are already grappling with the practical implications of AI in litigation, including a High Court decision striking out AI-generated court filings for procedural non-compliance.

Kenyan courts have also brought the human-rights implications of AI governance into sharper focus.  In February 2026, the High Court of Kenya certified as urgent a petition seeking to compel the Government to put in place a comprehensive AI policy and regulatory framework, on the argument that the unregulated deployment of high-risk AI systems threatens constitutional rights. The Court subsequently declined to restrain AI deployment but directed the matter to proceed to a full hearing, highlighting the rights-based dimension of the debate.

These rights-based concerns extend to AI-driven disinformation ahead of the 2027 general elections, an issue that has been raised in litigation and has formed part of the wider legislative and policy discussion on AI governance, reflecting broader public concerns about the technology’s social and democratic implications.

Taken together, the policy and legislative instruments under development in Kenya point to an AI governance landscape that is evolving rapidly but risks fragmentation. The priority for the next phase is less the generation of further instruments than the alignment of those already on the table into a coherent, coordinated and proportionate framework.

This article was written by Ariana Issaias, Partner, and Mumbi Kiragu, Associate, Bowmans