Kenya’s initial attempt to establish a dedicated legal framework for artificial intelligence (AI) has stalled at the committee stage. The Senate Standing Committee on Information, Communication, and Technology (ICT) rejected the Artificial Intelligence Bill, 2026, citing concerns that its punitive provisions and regulatory requirements were too burdensome for an ecosystem still finding its footing.   

Sponsored by Nominated Senator Karen Nyamu, the legislation proposed an ambitious administrative governance framework. At its core was the creation of the Office of the AI Commissioner, empowered to register AI systems, conduct audits, assess compliance, and handle public complaints. Complementing the Commissioner, the draft envisioned an AI Authority to set technical standards and operate regulatory sandboxes for supervised product testing, along with an Advisory Committee to guide government policy on emerging risks.   

However, the proposed enforcement mechanisms immediately attracted sharp public and parliamentary scrutiny. The draft bill introduced administrative fines of up to KES 5 million or a percentage of annual turnover, alongside criminal liability and prison sentences of up to two years for infractions such as harmful, non-consensual deepfakes. The Senate ICT Committee concluded that coupling strict licensing mandates with criminal exposure would actively discourage market participation, threatening to push local developers out of a market too young to absorb such high compliance costs.   

That assessment aligns with ground realities in East Africa’s tech hub. Kenya’s AI ecosystem is not anchored by multinational corporations with dedicated legal departments and compliance budgets. Instead, it is built by lean startups that adapt open-source models to tackle pressing local challenges in agriculture, healthcare, micro-lending, and education, operating on thin margins and with limited access to venture capital.  

Yet the committee’s decision leaves the core problem unresolved. Kenya remains without a dedicated statutory framework for AI. While the Data Protection Act, 2019 offers foundational rules on automated decision-making and profiling, it was never designed to address broader structural issues, including algorithmic bias, safety standards, public-sector AI deployments, or the proliferation of synthetic media.   

The rejected Bill rightly highlighted gaps in existing legal instruments, including the Science, Technology and Innovation Act and the Data Protection Act. Thus, the committee’s stance reflects a verdict on the specific design of Senator Nyamu’s draft rather than a dismissal of the need for oversight.   

Notably, the draft bill also sought to integrate devolved units into the national policy framework. By directing the AI Commissioner to advise county governments on deploying automated technologies across local healthcare and agricultural services, and by including Council of Governors representation on the Advisory Committee, the legislation acknowledged that counties are already deploying AI tools in service delivery without formal governance protocols.   

For now, two competing policy priorities are in tension:  

• Over-regulation: Prematurely imposing heavy regulatory burdens risks chilling tech investment and driving talent to markets with lighter, more flexible oversight.   

• Under-regulation: Leaving the ecosystem unmonitored exposes citizens to data privacy breaches, algorithmic discrimination, and deepfake manipulation, with limited legal recourse.   

The Senate Committee’s rejection reopens the national legislative debate. The challenge ahead for policymakers is to draft an adaptive, co-regulatory framework that enforces meaningful public safeguards without treating local software developers as operational liabilities.