South Africa’s draft national AI Policy was published in the Government Gazette on 10 April 2026, signalling a significant shift towards the formal regulation of artificial intelligence across all sectors of the economy. The policy sets out a comprehensive framework for AI governance, institutional oversight, ethical standards, and sector-specific strategies. Below are 10 key things that stakeholders, and particularly those in the financial services sector, should know.

The draft policy is built on six strategic pillars:

  • Capacity and talent development;
  • AI for inclusive growth and job creation;
  • Responsible governance;
  • Ethical and inclusive AI;
  • Cultural preservation and international integration; and
  • Human-centred deployment.

The policy proposes the establishment of several new institutional bodies to oversee AI governance. These include a National AI Commission (or National AI Office), an AI Ethics Board, an AI Regulatory Authority, an AI Ombudsperson Office, a National AI Safety Institute, an Integrated AI-Powered Monitoring Centre, and an AI Insurance Superfund. Together, these bodies are intended to coordinate strategy, monitor compliance, enforce governance, and provide redress mechanisms for individuals affected by AI-driven decisions.

The draft policy proposes the creation of an AI Insurance Superfund, modelled on the Road Accident Fund, to compensate individuals or entities harmed by AI-driven outcomes where liability is difficult to determine, especially for systems operating in ambiguous decision-making spaces. Firms in the financial services sector making use of AI in credit scoring, fraud detection or automated decision-making should monitor how this Superfund develops, as it may have direct implications for liability allocation and risk management.

The policy adopts a risk-based approach to AI governance, drawing some inspiration from the European Union AI Act. It proposes strict regulations for high-risk AI applications and categorises AI systems based on levels of potential harm. Financial systems are expressly identified as critical infrastructure requiring heightened protections. A National AI Safety Institute will be established to advance risk science and develop safety standards with international collaboration.

Explainability and transparency are central to the policy’s vision. All organisations making use of AI outputs will be expected to ensure that those outputs are sufficiently explainable and that any automated decisions are lawful, fair and challengeable. Automated systems must allow affected individuals to understand how decisions were made. Public-sector and high-risk AI systems will be subject to higher standards of transparency, including routine algorithm impact assessments, mandatory plain-language notifications when individuals are affected by automated decisions, and the publication of AI audit results.

Data protection is a key theme of the policy. The policy mandates alignment with the Protection of Personal Information Act, 2013 and the Promotion of Access to Information Act, 2000, and emphasises data protection by design, accountability, data minimisation and purpose limitation. Any cross-border transfers of data for AI processing will be required to comply with the National Policy on Data and Cloud (2024), with specific protocols envisaged for the financial services sector. Intellectual property rights are also addressed, with mandatory watermarking proposed to protect the rights of developers of large language models.

The policy prioritises ethical AI development and deployment. It identifies six key principles of responsible AI: fairness, reliability and safety, privacy and security, inclusiveness, transparency, and accountability. These principles will be embedded across the AI lifecycle. Mandatory gender and human rights impact assessments will be introduced for AI systems, particularly in high-risk domains. An independent oversight body will conduct certifications of high-stakes AI applications. The policy also addresses bias mitigation, requiring that AI systems be trained on diverse datasets and that algorithms be regularly tested for discriminatory outcomes.

Draft South African National Artificial Intelligence (AI) Policy

The policy places significant emphasis on human oversight of AI systems. It mandates a Human-in-the-Loop approach, requiring that critical AI decisions involve human oversight, especially in generative AI contexts. Safety-critical systems must include built-in safety mechanisms to remain under human control at all times. Decision-making architectures must prioritise human judgement over AI-generated decisions, particularly in essential government functions. For the financial services sector, this underscores the need for robust human review processes in AI-driven credit assessments, trading algorithms and customer-facing applications.

The policy envisages collaboration between financial regulators, with each regulator maintaining distinct responsibilities. These regulators include the South African Reserve Bank, the Financial Sector Conduct Authority, the Information Regulator, the Competition Commission, and the proposed new AI bodies. A National AI Regulatory Forum, coordinated by the Department of Communications and Digital Technologies, will bring together these regulators to coordinate oversight, standards and ethical guidance on AI in the financial sector and beyond.

Comments on the draft policy must be submitted by 10 June 2026. Financial services firms, industry bodies and legal advisors are encouraged to engage in the consultation process. In light of the proposed regulatory framework, financial services firms and regulators should begin preparing now by conducting internal AI audits, assessing current governance structures against the policy’s requirements, and evaluating the explainability and fairness of existing AI-driven systems. Proactive engagement with the consultation process will be critical in shaping a regulatory environment that balances innovation with consumer protection and systemic stability. As South Africa moves towards formalised AI regulation, early preparation will position financial institutions to adapt swiftly and maintain compliance whilst continuing to leverage AI’s transformative potential.