SCOTT TIMCKE AND HANANI HLOMANI: Protecting African elections with and from AI

In 2024 there are over 70 elections scheduled worldwide, with about 26 of them taking place in Africa, according to the Electoral Institute for Sustainable Democracy in Africa. These elections vary from presidential to local government elections.

That means more than half of the global population will have the chance to exercise their voting rights. However, malign uses of artificial intelligence (AI) threaten to undermine these democratic processes, particularly in Africa, where the voting populace has not yet been sufficiently educated about the potential dangers associated with the use of AI in democratic processes and the regulatory frameworks needed to govern AI in electoral processes are still lacking. 

However, Africa is not entirely unprepared to meet this challenge. Despite the absence of binding continental guidelines, many African countries have already laid the groundwork to regulate AI in elections through “soft laws” such as declarations and guiding principles. 

The foundational document is the 2002 Declaration on the Principles Governing Democratic Elections in Africa, adopted by the Organisation of African Unity, the predecessor to the AU. While not explicitly mentioning technology, this declaration establishes vital principles such as the necessity of impartial electoral bodies and respect for human rights and the rule of law — principles potentially affected both positively and negatively by the use of AI.

Building on this, the AU’s 2007 African Charter on Democracy, Elections and Governance includes specific provisions related to using technology to advance governance. In particular, the charter urges member states to employ information technology to promote political, economic and social development. However, it does not detail exactly how technology should be used in elections.

Critics charge that this lack of specificity renders the charter powerless to regulate more recent AI innovations such as micro-targeted political messaging. Yet the charter’s broad principles opposing discrimination and requiring transparent, impartial processes still provide an ethical framework applicable to AI.

Despite the charter’s limitations, individual African nations have taken steps to govern AI themselves. Many other African countries have introduced various policies touching on data protection, cybersecurity and e-governance. As it relates to data collection and processing, an essential part of AI use and development, data protection laws are arguably the most relevant policies and laws. 

To date, 35 African countries have enacted data protection laws while three are in the draft stage. What is common among the majority of data protection laws is that they often regulate how personal data — including voters’ information — can be processed, for example. This has created an expanding legal ecosystem to control AI even without binding continental standards.

AI governance regime

At the same time, the AU is ramping up efforts to build a comprehensive AI governance regime. The draft AU-AI continental strategy proposes increasing public awareness of AI’s societal impacts and regulating development of accountable AI systems. And the AU’s working group on AI plans further steps to govern AI on the continent.

With both grassroots progress and high-level plans under way, Africa has made promising strides towards managing AI’s double-edged potential. Yet much work remains as malign manipulators will surely attempt to exploit AI to undermine the continent’s democratic aspirations. 

To fortify Africa’s elections in 2024 and beyond AU member states should urgently ratify and implement the African Charter on Democracy, Elections and Governance’s core principles of transparency, impartiality and respect for human rights. The AU must establish a continental amendment stipulating precisely how AI can and cannot be used in electoral processes.

Binding guidelines should mandate transparency in AI systems used for political messaging and strictly limit micro-targeting to avoid manipulative personalisation. Strict audits must assess AI tools for bias that could disenfranchise women, minorities and other vulnerable groups. Electoral bodies should create specialised oversight capacities to monitor AI activities, aided by civil society watchdogs. Given that data protection authorities have been or are being established in the majority of African countries, this function might be delegated to them.

No doubt the road ahead will be challenging. But by leveraging its existing soft laws while expanding binding continental safeguards, Africa can harness AI’s potential for better governance while insulating democracy against its perils. With vigilance and collective action the continent’s citizens can exercise their votes freely and fairly even in AI’s emerging era.

• Dr Timcke is a senior research associate at Research ICT Africa, a research associate at the University of Johannesburg Centre for Social Change, and an affiliate of the Centre for Information, Technology & Public Life at the University of North Carolina. Hlomani is a research fellow with Research ICT Africa and a PhD candidate in tech and commercial law at the University of Cape Town.