National Artificial Intelligence Policy Framework (Draft)

This is a draft policy framework published for public consultation, not enacted legislation; references to it being "enacted" should be read as "published/adopted." South Africa's National Artificial Intelligence Policy Framework was published by the Department of Communications and Digital Technologies (DCDT) in October 2024, with a public consultation opened on 25 October 2024 (initially to 29 November 2024). It is expressly a first step, a foundational, principle- and risk-based roadmap intended to inform a fuller National AI Policy and, in due course, legislation. It has no binding legal force in its current form. The Framework's stated objective is to integrate AI technologies to drive economic growth, enhance societal wellbeing and position South Africa as a leader in AI innovation, while protecting rights and managing risk. It articulates a set of guiding principles centred on human-centred and ethical AI, transparency, accountability, fairness and inclusivity, and proposes a risk-based approach in which higher-impact (high-risk) AI systems attract greater scrutiny. The document is organised around a series of strategic pillars commonly summarised as including talent and skills development, digital and data infrastructure, research and innovation, standards and quality, public-sector AI adoption, ethics and safety, and multi-stakeholder governance and institutional coordination. It explicitly anchors AI governance in South Africa's existing legal framework, aligning with the Protection of Personal Information Act (POPIA) for the processing of personal data within AI systems, and signals the need for future statutory design to address gaps. Because it is a draft framework under consultation, it sets direction and invites comment rather than creating enforceable obligations; its provisions are proposals and principles, not duties. Readers should treat it as an early-stage policy statement signalling government intent and the likely shape of future regulation, with concrete legal effect contingent on the finalisation of a National AI Policy and any subsequent legislation. Note: the Instrument Type is recorded as "Guidance" reflecting its non-binding character; reviewers may prefer "Draft Bill" given its explicit draft/consultation status.

1. 1Published October 2024 by the DCDT as a draft framework, with public consultation opened 25 October 2024, intended as a first step toward a full National AI Policy.
2. 2Adopts a principle- and risk-based approach, with greater scrutiny for high-impact AI systems.
3. 3Centres human-centred, ethical AI with transparency, accountability, fairness and inclusivity as guiding principles.
4. 4Sets out strategic pillars including talent, digital and data infrastructure, research and innovation, standards, public-sector adoption and governance.
5. 5Aligns AI governance with the Protection of Personal Information Act (POPIA) for personal-data processing.
6. 6Signals the need for future statutory design and invites public comment rather than imposing binding obligations.