Kenya has introduced a structured risk-based approval system for (AI) Artificial Intelligence-powered medical devices, with foreign developers of the highest-risk software set to pay about KSh 3.2 million (approximately US$2,500) under new rules issued by the Pharmacy and Poisons Board.
The new framework classifies medical device software into four risk categories, with Category IV representing the most critical systems. These are tools whose failure could result in death or severe, irreversible harm to patients.
The regulation covers a wide range of technologies, including standalone AI diagnostic platforms, cloud-based clinical decision-making tools, and software integrated into medical equipment. Until now, these products had no dedicated approval pathway in Kenya.
Developers will also be required to submit a detailed Software Bill of Materials, effectively a full breakdown of every component within the system, including open-source code, third-party libraries, and version histories.
According to Dr Ahmed Mohamed, Acting CEO of the Pharmacy and Poisons Board, the framework makes a clear distinction between independent software and systems embedded in physical devices, a separation designed to better manage risks associated with AI and machine learning tools operating on their own.
Under the pricing structure, lower-risk software (Categories I and II), which mainly supports clinical decisions rather than driving them, will cost US$500 for foreign firms and US$250 for local developers. Category III products, which carry higher clinical risk, will be charged US$2,000 for foreign applicants and US$1,000 for local ones.
An additional administrative fee of US$300 applies to foreign companies across all categories, compared to US$100 for local manufacturers. All licences will remain valid for five years under the Health Products and Technologies Act.
Compared with global benchmarks, the fees remain relatively low. In the United States, regulatory clearance through the FDA can cost tens of thousands of dollars, while Europe’s conformity assessments under its Medical Device Regulation can run into hundreds of thousands of euros.
The pricing strategy appears to position Kenya as a more accessible entry point for health-tech firms targeting Africa’s growing digital health market, particularly as companies look beyond saturated Western markets.
The framework also tightens safety and transparency requirements. Software distributed physically must include clear user instructions, while online tools must display essential information such as intended use and safety warnings within their interface. Developers are also expected to maintain strong version control systems to ensure traceability and enable corrective action when needed.
AI medical device manufacturers must further disclose details on training data, model design, bias mitigation strategies, and clinical performance indicators such as sensitivity and specificity.
Also Read: Kenya Power to Procure Meters Under KSh 19bn Off-Grid Solar Drive
For adaptive AI systems that continue learning after deployment, firms must include safeguards such as anomaly detection, rollback functions, and full tracking of how real-world data influences outcomes. They will also be required to submit annual performance reports throughout the product’s lifecycle.
The new rules are aligned with Kenya’s wider digital governance agenda, including the Digital Health Act 2023, the National Cybersecurity Strategy to 2027, and the Kenya AI Strategy extending to 2030. The classification system also mirrors international standards set by the International Medical Device Regulators Forum, with applications managed through an online tracking portal.
