The Artificial Intelligence Bill, 2026 recently tabled in Parliament, marks a significant and commendable milestone in Kenya’s journey towards robust AI governance. The Bill reflects a more adaptive and development-oriented model, balancing oversight with innovation, capacity building, and ecosystem growth.
While the Bill is a bold and progressive initiative, these certain areas could benefit from further clarity to strengthen its effectiveness:
1. Definition of Artificial Intelligence
The Bill ought to clearly distinguish AI systems from traditional software or purely rule-based programs, excluding software systems that function solely on fixed rules established by human operators and lack the capacity to learn, adapt, or produce outputs autonomously.
2. Scope of Application
It remains unclear whether the Bill applies to deployers or AI systems that are established in foreign jurisdictions but are largely used within Kenya’s territorial jurisdiction.
3. Roles of Distributors and Third-Party Providers
Clear definitions and obligations of distributors and third-party providers of AI Systems will help ensure accountability across the AI supply chain.
4. Risk Classification
The criteria or threshold used to assess the level of risks posed by an AI System, to classify the system as either unacceptable risk, High Risk, limited risk or minimal risk is not clear.
5. Prohibited AI Practices
The Bill currently offers no comparable prohibition, and could benefit from specifying prohibited AI practices to provide clear boundaries in a fast developing technological arena.
This is particularly important in areas with high ethical or societal impact. Such provisions could draw inspiration from frameworks like Article 5 of the EU AI Act, which identifies practices considered inherently unacceptable, including mass biometric surveillance and manipulative or deceptive uses of AI.
Considering the Bill is still at a preliminary stage, it will be especially interesting to see how this framework matures in practice, the accompanying regulations that will guide enforcement, and how it interacts with more established regimes like the EU’s in cross-border AI deployment and compliance.
Leave a Reply