South Africa Edges Closer to National AI Framework: Draft Policy Heads to Cabinet, Public Input Set for March 2026
South Africa is accelerating its journey toward structured artificial intelligence governance, with the Draft National AI Policy now advancing through the critical Cabinet approval stage. This development marks a major progression from broad strategic discussions to actionable regulatory steps, as confirmed during a key briefing to Parliament's Portfolio Committee on Communications and Digital Technologies on 24 February 2026 by the Department of Communications and Digital Technologies (DCDT).
Having successfully cleared the Socio-Economic Impact Assessment System (SEIAS) certification and secured concurrence from all relevant Director-General clusters, the draft policy enjoys strong inter-governmental backing. It is currently navigating Cabinet processes, with expectations that it will be published—gazetted—in the Government Gazette for a 60-day public consultation period starting in March 2026. Final adoption of the policy is targeted for the 2026/2027 financial year, paving the way for sector-specific strategies and implementation measures from 2027/2028 onward.
The government's approach deliberately avoids establishing a centralized, standalone AI regulator—in contrast to models like the EU AI Act. Instead, it embraces a sector-specific, multi-regulator framework that integrates AI oversight into existing bodies such as ICASA (for communications and digital infrastructure), financial regulators, health authorities, and others. This phased, tailored model recognizes that AI risks and applications vary widely across industries like healthcare, finance, telecommunications, and public services.
At the heart of the Draft National AI Policy are five core pillars designed to balance innovation with responsibility and inclusivity:
Capacity and talent development — Prioritizing national AI skills through education, training, industry partnerships, expanded digital infrastructure (including more compute resources like GPUs), and support for SMEs to foster local innovation and economic participation.
Responsible AI governance — Introducing safeguards against risks such as data misuse, cybersecurity vulnerabilities, misinformation, deepfakes, and other harms, while enforcing clear accountability.
Ethical and inclusive AI — Addressing bias mitigation by advocating for training on representative South African datasets to prevent "imported bias" from Global North models, alongside national ethical guidelines and developer accountability for adverse outcomes.
Cultural preservation and global integration — Leveraging AI to digitize and protect indigenous languages (with specific mention of Khoi and San languages) and traditional knowledge systems, while bolstering international collaboration to keep South Africa competitive.
Human-centred deployment — Rejecting opaque "black box" systems in high-impact areas and insisting on human oversight, explainability, and organizational responsibility for AI-driven decisions—even in automated contexts.
The policy positions AI as a driver of inclusive economic growth, job creation, improved public services, and global competitiveness, while emphasizing equitable distribution of benefits and risks across society and generations. It responds to concerns about concentrated AI power and aims to empower local talent, protect privacy and security, and align technological advancement with South African values.
With public consultation imminent, organizations, businesses, civil society, and citizens are urged to prepare: monitor the gazetting notice, engage actively during the 60-day comment window to shape sector-specific rules, and conduct internal audits of current AI systems. This includes mapping high-impact deployments, reviewing data practices, assessing explainability, ensuring compliance with laws like POPIA (Protection of Personal Information Act), and aligning with governance standards such as King V.
South Africa's measured, context-sensitive path reflects lessons from global peers like the UK and India—prioritizing embedded oversight over heavy new bureaucracy—while addressing unique local priorities like cultural heritage, bias in diverse demographics, and bridging digital divides. As 2026 unfolds, this evolving framework could position the country as a thoughtful leader in Africa's AI landscape, turning potential risks into opportunities for sustainable, people-first innovation.