CIPESA calls for stronger safeguards in Kenya’s draft AI policy

The Collaboration on International ICT Policy for East and Southern Africa (CIPESA) has submitted recommendations on Kenya’s draft Artificial Intelligence (AI) and Other Emerging Technologies Policy, calling for a more coherent and rights-based approach to AI governance.

CIPESA argues that Kenya is developing several AI-related instruments at the same time, including a national AI strategy, a proposed AI Bill and the draft policy. The organisation recommends aligning these processes to avoid overlapping responsibilities and conflicting rules. It specifically points to a provision in the draft policy proposing a new AI governance law despite a similar bill already before the Senate.

The submission calls for stronger legal safeguards around the use of AI, particularly where systems could affect fundamental rights. CIPESA recommends mandatory human rights and gender impact assessments for high-risk AI systems, with risks identified and addressed before deployment rather than only after harm occurs.

The organisation also raises concerns about the independence and authority of the proposed AI Council. It recommends giving the oversight body clear powers to audit AI systems, obtain relevant information, enforce compliance and report directly to Parliament. The submission also calls for clearer rules on responsibility and liability so that people affected by automated decisions have effective ways to challenge decisions and seek redress.

CIPESA highlights additional concerns around the use of AI in Kenya, including discrimination, surveillance, privacy, freedom of expression, political manipulation and AI-generated disinformation. It recommends stronger oversight of AI used in elections, political advertising and content moderation, as well as explicit safeguards against technology-facilitated gender-based violence.

The submission also addresses the environmental and labour impacts of AI. CIPESA calls for greater accountability for the energy and water use, emissions and electronic waste associated with AI infrastructure, including independent verification of environmental disclosures. It also recommends stronger protections for workers involved in AI systems, including data annotators and content moderators.

According to CIPESA, AI systems used in African contexts can also produce biased results when training data and moderation tools do not adequately represent local populations or languages. The organisation therefore calls for stronger local data ecosystems and greater support for African languages in AI development and content moderation.

Finally, CIPESA recommends more meaningful participation by civil society in AI governance structures and clearer public processes for understanding and contributing to policy. It also calls for regular review of the policy as AI technologies and their risks continue to evolve.