On April 8, 2024, the Kenya Bureau of Standards (KEBS) published the draft Information Technology - Artificial Intelligence - Code of Practice for AI Applications (the Code) for public consultation. In particular, the Code provides a set of recommendations intended to help organizations develop, provide, or use artificial intelligence (AI) systems responsibly in pursuing their objectives. To this end, the Code outlines:

• approaches to establish trust in AI systems through transparency, explainability, and controllability;
• engineering pitfalls and typically associated threats and risks to AI systems, along with possible mitigation techniques and methods; and
• approaches to assess and achieve availability, resiliency, reliability, accuracy, safety, security, and privacy of AI systems.

Stakeholder responsibilities

Additionally, the Code discusses the roles played by various AI stakeholders in stages of the AI system life cycle and their corresponding responsibilities. Recommending that organizations involved in designing, developing, testing, and deploying products or services utilizing AI systems should:

• specify the characteristics of the AI application and disclose the data used in the model's creation, including the data sources, data providers, and any partnerships involved;
• assess and manage the risks of the AI system through an established risk management system; and
• specify the stages of the AI system lifecycle in which model building, application, and updates are reviewed.

Further, the Code highlights the responsibilities of AI application providers, defined as organizations or entities that provide products or services that use AI systems. These responsibilities include:

• managing the operational analytics of the AI application and ensuring they are monitored continuously;
• overseeing how the AI application is built, applied, and updated, including the training and programming of the AI model;
• assessing and addressing the impacts of the AI application on its customers, users, and the broader community; and
• defining any boundaries for the responsible use of the AI application and ensuring these are incorporated into legal requirements, including software licenses.

Comments on the Code can be submitted to the Managing Director of the KEBS through this form until June 13, 2024.

You can read the announcement here and the Code here.