AI Safety Regulation: Lessons for Africa

Artificial Intelligence (AI) has the potential to alter various facets of daily life in Africa with applications spanning diverse areas such as employment, education, health and justice.

It holds immense potential for transformative impact in the area of science, clean energy, biodiversity and climate action. While AI offers unparalleled opportunities, some argue that the potential risks require detailed regulations, including potentially on an African-wide basis. Other blog posts in this edition of Eye on Emerging Markets highlight the recent Global AI Summit that took place at Bletchley Park in the United Kingdom as well as President Biden's recent Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence. More and more African countries are adopting new laws and regulations relating to AI, generally with the aim of fostering responsible development. For example:

• Kenya, Rwanda and Nigeria endorsed the recent Bletchley Declaration;
• Kenya has proposed the Kenya Robotics and Artificial Intelligence Society Bill, 2023 which would establish a "Kenya Robotics and Artificial Intelligence Society" to regulate, promote and facilitate the activities of robotics and AI practitioners within the country;
• Rwanda's "National AI Policy" prioritises trustworthy AI adoption in the public sector and beneficial AI adoption in the private sector and its Utilities Regulatory Authority is also reported to be in the final stages of developing practical, ethical guidelines for the use of AI;
• Nigeria's National Information Technology Development Agency is progressing the country's "National Artificial Intelligence Policy" that will provide direction on its strategy to mitigate AI risk;
• Egypt has a charter for responsible AI that sets out various guidelines on the ethical and responsible development, deployment, use and management of AI;
• Tunisia's National Agency for the Protection of Personal Data has issued recommendations on ethical AI development.

However, though African countries are now striving towards enacting national comprehensive regulatory frameworks on AI safety, they may also seek cost efficiencies for both market participants and regulators by adopting uniform regulations across multiple countries.  Whilst multilateral agreements are challenging to reach, examples such as the African Continental Free Trade Agreement and the OHADA legal harmonisation system adopted by seventeen West and Central African countries to assist their economic and legal integration many years ago show that it can be done.

In addition, as with other countries and regions around the world, regulators and legislators have to balance the desire for regulation with the dangers of over-regulating an innovative and fast moving industry. As such, African jurisdictions have been paying close attention to the safety standards and principles being adopted by other countries and regions, in particular with respect to:

• whether to categorise AI based on potential risks and to focus oversight on high-risk categories/uses;
• the EU risk-based approach where the obligations for an AI system are proportionate to the level of risk that it poses;
• requiring developers to share critical information with the government and implementing rigorous red-team testing;
• prioritising privacy-preserving techniques as well as developing guidelines for evaluating their effectiveness in protecting citizens' privacy;
• establishing or collaborating with an AI safety body/institute;
• investing in AI safety training programmes; and
• the need for regulation to be proportionate and not overly costly or burdensome for market participants.

African regulators have also acknowledged that regulations with respect to incident reporting for serious AI-related incidents will also be key for ensuring continuous improvement. While considering the specifics of their context, African nations can adapt these elements to fortify their AI safety frameworks whilst still encouraging investment and innovation.

Striking a balance between fostering innovation and safeguarding against potential risks is a critical challenge that African policymakers need to make. As Africa navigates the complex landscape of AI safety regulation, international cooperation and the sharing of best practices become paramount. The region has the opportunity to shape ethical and responsible development throughout the continent that will drive innovation, whilst prioritising human rights, transparency and environmental sustainability. Collaborative efforts will not only safeguard against potential harms but also ensure that AI fulfils its transformative potential for the continent's future prosperity and well-being.