Jump to content

South Africa AI Law

From AI Law Wiki

AI-generated text. This page was generated using artificial intelligence.

South Africa AI law combines existing data-protection rules with a national AI-policy process. The Cabinet withdrew its 2026 draft National Artificial Intelligence Policy in June to rework it; the withdrawn draft is not an operative AI law or current government policy.[1] Source review: September 27, 2026.

Existing law

The Protection of Personal Information Act 4 of 2013 (POPIA) governs the processing of personal information by covered public and private bodies, including processing performed with AI. Section 71 addresses decisions based solely on automated processing intended to profile a person where the decision has legal consequences or substantially affects that person. Its restriction has specified exceptions for contract-related decisions and decisions governed by law or a code of conduct with safeguards; where a contract exception depends on protective measures, the Act requires an opportunity to make representations and information about the processing logic.[2] The Information Regulator describes prior-authorisation requirements for specified higher-risk processing under section 57, which are separate from section 71's automated-decision rule.[3]

National AI policy

Cabinet approved publication of a draft national AI policy for comment in March 2026. Its proposed pillars included responsible governance, ethical and inclusive AI, skills, and human-centred deployment.[4] Cabinet approved withdrawal of that draft at its June 3 meeting so the policy could be reworked. The official withdrawal notice appeared in Government Gazette 54840 on June 12.[1][5]

See October 29 reporting on the revised policy’s drafting timetable.

Western Cape infrastructure cooperation

On October 1, 2026, the Western Cape Department of Infrastructure announced a five-year memorandum of understanding with the Geneva-based AI for Developing Countries Forum (AIFOD). The department describes cooperation on responsible AI in infrastructure planning, governance and delivery. Announced work includes project preparation and monitoring, human-centred accountability, research and capacity building. It will explore AIFOD’s Future AI Investment Responsibility (FAIR) certification and governance model for departmental investments; the announcement does not establish a general certification duty. A Cape Town summit is planned for April 2027.[6]

This account is based on the department’s public announcement; the memorandum’s text has not been reviewed. See Governance for related public-sector AI oversight coverage.

International proposals

At the September 13, 2026 BRICS leaders' summit, President Cyril Ramaphosa proposed an international mechanism for independent scientific evaluation of AI, with human-control principles and incident reporting. This was a South African proposal, not an adopted BRICS instrument.[7] South Africa's UN representative also called for global AI-governance guardrails in a September 22 government interview.[8] These positions should be distinguished from the existing international framework inventory.

On October 8, 2026, Parliament reported that South Africa’s delegation to the 153rd Inter-Parliamentary Union Assembly in Arusha, Tanzania, had called that Monday for a binding international instrument on military AI. Delegate Doris Mpapane urged safeguards for meaningful human control, inclusive multilateral rulemaking, and parliamentary scrutiny of procurement, development and deployment to support transparency, accountability and compliance with international law. These are the delegation’s proposals for international rules.[9]

References