Why You Need To Engage And Comment On The Draft South Africa National AI Policy

On 10 April 2026, the Minister of Communications and Digital Technologies published the Draft South Africa National Artificial Intelligence Policy (“Draft National AI Policy”). The Draft National AI Policy is based on the South Africa National Artificial Intelligence Policy Framework published in August 2024 and seeks to promote responsible innovation and the safe and secure adoption of artificial intelligence (“AI”) technologies. Its stated objectives include safeguarding fundamental rights, driving social and economic transformation and ensuring alignment with international norms and standards.

Structure and key pillars of the Draft National AI Policy

The Draft National AI Policy is structured around six strategic pillars, each supported by strategic building blocks:

• Capacity and talent development

This pillar focuses on integrating AI education from primary through to tertiary level, reskilling and upskilling workers in sectors most affected by AI, and establishing community-based AI education centres. A key aspect of this pillar is investment in digital infrastructure in order to promote AI sovereignty.

• AI for inclusive growth and job creation

To accelerate AI adoption, the Draft National AI Policy proposes establishing AI research centres and AI accelerators supported by regulatory measures and targeted financial incentives such as tax credits and an AI Innovation Fund, to promote economic growth and productivity.

• Responsible governance

This pillar covers cybersecurity measures to protect AI systems, risk management frameworks that are sensitive to South Africa’s socio-economic conditions, data sharing protocols, and protections for children against manipulative AI practices. The Draft National AI Policy aligns AI privacy measures with the Protection of Personal Information Act, 2013 (“POPIA”) and identifies intellectual property as a critical governance area, particularly in relation to the use of copyrighted materials to train large language models.

• Ethical and inclusive AI

Bias is a significant risk associated with the use of AI. To mitigate this risk, the Draft National AI Policy introduces six guiding principles for responsible AI: (i) fairness, (ii) reliability and safety, (iii) privacy and security, (iv) inclusiveness, (v) transparency, and (vi) accountability. A targeted, risk-based regulatory approach is proposed, under which AI systems are classified by risk level, with strict protections for critical infrastructure and high-risk use cases.

• Cultural preservation and international integration

AI is to be leveraged for real-time translation across South Africa’s 12 official languages, as well as the digitisation and preservation of indigenous culture. The Draft National AI Policy seeks to align with the African philosophy of Ubuntu, while harmonising with international AI frameworks.

• Human-centred deployment

The Draft National AI Policy mandates a Human-in-the-Loop approach for critical decision-making, particularly in the context of generative AI. Public-sector AI systems will be subject to heightened standards of explainability and accountability, with pilot projects planned in healthcare, education and urban planning.

Proposed AI regulatory architecture

In addition to existing regulators that will have a key role in enforcing AI regulation in South Africa, such as the Information Regulator, the Independent Communications Authority of South Africa (“ICASA”), and the Competition Commission, the Draft National AI Policy proposes the establishment of several new entities to regulate AI in South Africa. These include:

• a National AI Commission Office, responsible for coordinating policy implementation and facilitating stakeholder engagement;
• an AI Ethics Board, tasked with overseeing ethical governance, including issues of bias, privacy and fairness;
• an AI Regulatory Authority, mandated to monitor compliance, conduct audits and issue certifications;
• an AI Ombudsperson Office, providing individuals with a mechanism to challenge AI-driven decisions; and
• an AI Insurance Superfund, that will process claims and compensate individuals harmed by AI-driven outcomes.

Implementation Timeline

The Department of Communications and Digital Technologies (“DCDT”) plans to implement the Draft National AI Policy over a three-year period. Year 1 will focus on finalising the policy and publishing draft regulatory requirements for addressing unacceptable AI risks. Year 2 will see the publication of National AI Policy Guidelines, implementation of rules for high-risk AI use cases, development of draft regulations for medium- and low-risk AI use cases, and the commencement of sector specific and institutional design strategies. Year 3 will involve implementation of the remaining policy interventions and updating the framework to address emerging AI trends.

Why stakeholder engagement matters

At first glance, the Draft National AI policy may appear abstract, aspirational and somewhat removed from day‑to‑day business realities. This is deliberate, as at this stage, the DCDT is seeking extensive input from local and international experts, private bodies, and interest groups before finalising the policy.

An important aspect of the Draft National AI Policy is its emphasis on sector‑specific considerations, reflecting the reality that the impact of AI varies across industries and that certain sectors have a greater interest in how the policy is ultimately shaped.

• Financial services providers and fintechs

The Draft National AI Policy is likely to have a direct impact on AI use cases related to credit scoring, fraud detection, and underwriting purposes being classified as high-risk, triggering obligations around explainability and auditability. This could materially affect financial model design and governance. Stakeholder engagement is encouraged to shape how high-risk financial AI is defined and to ensure alignment with existing FSCA and SARB frameworks.

• Technology companies, AI developers, SaaS providers, and platforms

The Draft National AI Policy highlights the role that technology companies play in AI governance by distinguishing between developers, deployers and integrators, with potential shared liability across the value chain. New obligations relating to ethical standards and transparency when deploying AI could significantly affect product design and technology business models. Key issues that technology companies and developers need to consider include the use of open‑source models and training‑data, data ownership, and ICT infrastructure and cloud services regulations.

• Telecommunications, media and digital platforms

Telecommunications operators and digital platforms are directly implicated by the Draft National AI Policy’s proposal to expand ICASA’s mandate into AI oversight. Stakeholders should engage on the AI regulatory overlap between ICASA, the Information Regulator and the Competition Commission, as well as the intellectual property protections and governance issues relating to AI‑generated content.

• Healthcare, life sciences and medical technology

In healthcare, the use of AI for diagnostic and decision‑making is likely to be treated as high‑risk, given patient safety and liability considerations. Healthcare institutions are already subject to POPIA obligations and strict duties of confidentiality in relation to health information, and the Draft National AI Policy introduces additional requirements that will need to account for distinctions between clinical and administrative use of AI, appropriate levels of human oversight when using AI in the healthcare sector, and the rules governing medical research.

• Energy, mining, manufacturing and logistics

AI systems used in operational environments may be classified as critical or high‑risk, with significant implications for planning, risk management, governance and security. This classification could introduce heightened governance, oversight and security expectations for affected entities. Stakeholder engagement is therefore important to shape how risk is defined and how implementation timelines and compliance obligations are realistically structured.

• Employers, HR technology providers and labour‑intensive industries

The use of AI for recruitment, performance monitoring and workforce management raises labour-law issues. The Draft National AI Policy recognises job displacement as a major AI-related risk and proposes mitigation measures. As such, employers should make submissions regarding the acceptable use of AI for recruitment and performance management, reskilling obligations, and alignment with existing labour legislation.

The Draft National AI Policy is deliberately framed using open-ended, principles-based provisions, creating meaningful opportunities for stakeholders to influence its final form. Interested parties have until 10 June 2026, 16h00 to submit their comments and written representations.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.