Authorities advances sector‑primarily based AI regulation as public session nears
In short
South Africa’s Draft Nationwide AI Coverage has formally entered the Cupboard approval course of, signalling a decisive shift from excessive‑stage ideas to concrete regulatory improvement. The Coverage is anticipated to be gazetted for a 60‑day public session in March 2026, with finalisation focused for the 2026/2027 monetary yr. Authorities has opted for a sector‑particular, multi‑regulator mannequin slightly than a single AI regulator, which means AI governance will probably be embedded inside current supervisory frameworks. 5 core pillars – abilities capability, accountable governance, moral and inclusive AI, cultural preservation and human‑centred deployment – information the Coverage. Organisations ought to assess present AI deployments, governance constructions and excessive‑affect programs forward of elevated oversight and future sector‑particular regulatory devices.
In additional element
The Highway to Gazetting: Timelines and Procedures
On 24 February 2026, the Division of Communications and Digital Applied sciences (DCDT) briefed Parliament on the progress of the Draft Nationwide AI coverage (“Draft AI Coverage“). The DCDT’s briefing confirmed that the Draft AI Coverage has efficiently cleared the Socio-Financial Affect Evaluation System certification and achieved concurrence throughout all Director-Common clusters. It is a essential administrative hurdle; it suggests the Coverage has broad inter-departmental assist and is prepared for public scrutiny.
The Draft AI Coverage is now progressing via Cupboard approval and is anticipated to be gazetted for a 60-day public session interval in March 2026. Finalisation is anticipated throughout the 2026/2027 monetary yr, with sector-specific methods and supporting regulatory measures anticipated to comply with from 2027/2028.
Strategic Context of the Draft AI Coverage
A key focus of the Parliamentary briefing was on the advantages and dangers of AI and the way every of those needs to be extra evenly distributed throughout society. Authorities highlighted considerations in regards to the present focus of AI capabilities and emphasised the necessity to guarantee long-term advantages for each present and future generations.
AI was additionally framed as a software to assist inclusive financial development. The Draft AI Coverage goals to strengthen Authorities’s means to control and undertake AI responsibly, whereas encouraging native innovation, supporting job creation and enhancing entry to AI abilities, infrastructure and providers. These priorities sit alongside efforts to reinforce South Africa’s international competitiveness and worldwide collaboration.
5 Core AI Coverage Pillars
It was in opposition to this backdrop that the next 5 core coverage pillars underpinning the Draft AI Coverage had been launched by the Authorities throughout the Parliamentary briefing.
- Capability and expertise improvement
The DCDT highlighted the necessity to construct nationwide AI abilities via schooling, coaching and nearer collaboration with business. This will probably be supported by improved digital infrastructure, together with elevated compute capability (equivalent to graphics processing models) and higher connectivity, to assist native innovation, SMEs and broader financial development.
- Accountable AI governance
The Draft AI Coverage proposes sensible safeguards to handle security, safety and privateness dangers related to AI. These embrace dangers equivalent to knowledge misuse, cybersecurity threats, misinformation and deepfakes, with the intention of guaranteeing that AI programs are deployed with clear accountability and don’t trigger hurt.
Equity and bias mitigation had been central to the dialogue. Authorities highlighted the significance of coaching AI programs on consultant native datasets to keep away from imported bias (i.e., discriminatory outcomes when AI fashions educated on World North datasets are utilized to South African demographics). The event of nationwide moral pointers and mechanisms to carry builders accountable for dangerous outcomes was additionally emphasised.
- Cultural preservation and international integration
The Draft AI Coverage goals to make use of AI to protect indigenous languages and data programs whereas strengthening worldwide collaboration. On this regard, the Koi and Satsang languages had been particularly talked about with the intention of AI digitising them. Authorities additionally famous the significance of defending cultural property whereas guaranteeing South Africa stays globally aggressive.
The Draft AI Coverage rejects opaque “black field” deployment in high-impact contexts and emphasises accountability when AI programs trigger hurt. Which means Which means organisations stay answerable for the outcomes of AI-driven choices, whatever the stage of automation concerned.
Implementation mannequin and sector-specific regulation
The DCDT additionally addressed the supposed implementation plan of the Draft AI Coverage in follow. Authorities indicated {that a} phased method will probably be adopted, recognising that AI deployment and danger profiles differ considerably throughout sectors. AI in healthcare, for instance, presents distinct moral and security issues when in comparison with AI utilized in monetary providers, telecommunications or public administration.
Because of this, the Draft AI Coverage is anticipated to function as an overarching framework, with sector-specific methods developed thereafter and ruled inside current sector-specific regulatory frameworks. That is an method favoured in different international locations equivalent to the UK and India, and indicators one other departure from the EU AI Act’s complete regulatory method. These methods will probably be tailor-made to the regulatory realities and danger exposures of particular person industries.
Maybe probably the most important structural revelation from the DCDT is the choice to not create a single AI Regulator. As an alternative, oversight will probably be distributed amongst current authorities – ICASA was particularly talked about throughout the briefing as one of many authorities anticipated to play a task in relation to digital infrastructure and communications elements of AI deployment.
This multi-regulator mannequin represents a coordinated oversight method slightly than the creation of a centralised AI regulator. Because of this, AI governance is prone to intersect with current regulatory obligations – equivalent to these referring to conduct, danger administration, knowledge safety, and cybersecurity – embedding AI accountability inside established supervisory frameworks slightly than introducing it via a standalone regime.
The best way ahead
With the Draft AI Coverage anticipated to enter public session shortly, organisations ought to deal with 2026 as a interval to arrange strategically.
Within the quick time period, establishments ought to monitor the gazetting course of and think about taking part within the 60-day public remark interval. Early engagement might affect how sector-specific methods are finally framed, significantly in relation to explainability and supervisory oversight.
At an operational stage, establishments ought to conduct a complete inner assessment of current AI deployments. This consists of figuring out high-impact programs, mapping knowledge flows, assessing mannequin explainability mechanisms and evaluating alignment with current governance frameworks and suggestions (equivalent to these below King V) and obligations below current regulation, equivalent to POPIA, prudential requirements and conduct regulation. As governance expectations evolve, regulators are prone to assess AI via the lens of current accountability frameworks slightly than in isolation.
The 2026 Draft Nationwide AI Coverage represents an effort by the South African Authorities to meet up with international regulatory developments whereas addressing native socio-economic realities. The transition from the present open regulatory setting to a sector-specific framework will probably be complicated, nevertheless it additionally provides a possibility for companies to construct better belief with their prospects via proactive engagement, because the method in direction of governing AI in South Africa is mapped.
* * * * *
Despina Lazanakis, Trainee Solicitor, has contributed to this authorized replace.
Authorities advances sector‑primarily based AI regulation as public session nears
In short
South Africa’s Draft Nationwide AI Coverage has formally entered the Cupboard approval course of, signalling a decisive shift from excessive‑stage ideas to concrete regulatory improvement. The Coverage is anticipated to be gazetted for a 60‑day public session in March 2026, with finalisation focused for the 2026/2027 monetary yr. Authorities has opted for a sector‑particular, multi‑regulator mannequin slightly than a single AI regulator, which means AI governance will probably be embedded inside current supervisory frameworks. 5 core pillars – abilities capability, accountable governance, moral and inclusive AI, cultural preservation and human‑centred deployment – information the Coverage. Organisations ought to assess present AI deployments, governance constructions and excessive‑affect programs forward of elevated oversight and future sector‑particular regulatory devices.
In additional element
The Highway to Gazetting: Timelines and Procedures
On 24 February 2026, the Division of Communications and Digital Applied sciences (DCDT) briefed Parliament on the progress of the Draft Nationwide AI coverage (“Draft AI Coverage“). The DCDT’s briefing confirmed that the Draft AI Coverage has efficiently cleared the Socio-Financial Affect Evaluation System certification and achieved concurrence throughout all Director-Common clusters. It is a essential administrative hurdle; it suggests the Coverage has broad inter-departmental assist and is prepared for public scrutiny.
The Draft AI Coverage is now progressing via Cupboard approval and is anticipated to be gazetted for a 60-day public session interval in March 2026. Finalisation is anticipated throughout the 2026/2027 monetary yr, with sector-specific methods and supporting regulatory measures anticipated to comply with from 2027/2028.
Strategic Context of the Draft AI Coverage
A key focus of the Parliamentary briefing was on the advantages and dangers of AI and the way every of those needs to be extra evenly distributed throughout society. Authorities highlighted considerations in regards to the present focus of AI capabilities and emphasised the necessity to guarantee long-term advantages for each present and future generations.
AI was additionally framed as a software to assist inclusive financial development. The Draft AI Coverage goals to strengthen Authorities’s means to control and undertake AI responsibly, whereas encouraging native innovation, supporting job creation and enhancing entry to AI abilities, infrastructure and providers. These priorities sit alongside efforts to reinforce South Africa’s international competitiveness and worldwide collaboration.
5 Core AI Coverage Pillars
It was in opposition to this backdrop that the next 5 core coverage pillars underpinning the Draft AI Coverage had been launched by the Authorities throughout the Parliamentary briefing.
- Capability and expertise improvement
The DCDT highlighted the necessity to construct nationwide AI abilities via schooling, coaching and nearer collaboration with business. This will probably be supported by improved digital infrastructure, together with elevated compute capability (equivalent to graphics processing models) and higher connectivity, to assist native innovation, SMEs and broader financial development.
- Accountable AI governance
The Draft AI Coverage proposes sensible safeguards to handle security, safety and privateness dangers related to AI. These embrace dangers equivalent to knowledge misuse, cybersecurity threats, misinformation and deepfakes, with the intention of guaranteeing that AI programs are deployed with clear accountability and don’t trigger hurt.
Equity and bias mitigation had been central to the dialogue. Authorities highlighted the significance of coaching AI programs on consultant native datasets to keep away from imported bias (i.e., discriminatory outcomes when AI fashions educated on World North datasets are utilized to South African demographics). The event of nationwide moral pointers and mechanisms to carry builders accountable for dangerous outcomes was additionally emphasised.
- Cultural preservation and international integration
The Draft AI Coverage goals to make use of AI to protect indigenous languages and data programs whereas strengthening worldwide collaboration. On this regard, the Koi and Satsang languages had been particularly talked about with the intention of AI digitising them. Authorities additionally famous the significance of defending cultural property whereas guaranteeing South Africa stays globally aggressive.
The Draft AI Coverage rejects opaque “black field” deployment in high-impact contexts and emphasises accountability when AI programs trigger hurt. Which means Which means organisations stay answerable for the outcomes of AI-driven choices, whatever the stage of automation concerned.
Implementation mannequin and sector-specific regulation
The DCDT additionally addressed the supposed implementation plan of the Draft AI Coverage in follow. Authorities indicated {that a} phased method will probably be adopted, recognising that AI deployment and danger profiles differ considerably throughout sectors. AI in healthcare, for instance, presents distinct moral and security issues when in comparison with AI utilized in monetary providers, telecommunications or public administration.
Because of this, the Draft AI Coverage is anticipated to function as an overarching framework, with sector-specific methods developed thereafter and ruled inside current sector-specific regulatory frameworks. That is an method favoured in different international locations equivalent to the UK and India, and indicators one other departure from the EU AI Act’s complete regulatory method. These methods will probably be tailor-made to the regulatory realities and danger exposures of particular person industries.
Maybe probably the most important structural revelation from the DCDT is the choice to not create a single AI Regulator. As an alternative, oversight will probably be distributed amongst current authorities – ICASA was particularly talked about throughout the briefing as one of many authorities anticipated to play a task in relation to digital infrastructure and communications elements of AI deployment.
This multi-regulator mannequin represents a coordinated oversight method slightly than the creation of a centralised AI regulator. Because of this, AI governance is prone to intersect with current regulatory obligations – equivalent to these referring to conduct, danger administration, knowledge safety, and cybersecurity – embedding AI accountability inside established supervisory frameworks slightly than introducing it via a standalone regime.
The best way ahead
With the Draft AI Coverage anticipated to enter public session shortly, organisations ought to deal with 2026 as a interval to arrange strategically.
Within the quick time period, establishments ought to monitor the gazetting course of and think about taking part within the 60-day public remark interval. Early engagement might affect how sector-specific methods are finally framed, significantly in relation to explainability and supervisory oversight.
At an operational stage, establishments ought to conduct a complete inner assessment of current AI deployments. This consists of figuring out high-impact programs, mapping knowledge flows, assessing mannequin explainability mechanisms and evaluating alignment with current governance frameworks and suggestions (equivalent to these below King V) and obligations below current regulation, equivalent to POPIA, prudential requirements and conduct regulation. As governance expectations evolve, regulators are prone to assess AI via the lens of current accountability frameworks slightly than in isolation.
The 2026 Draft Nationwide AI Coverage represents an effort by the South African Authorities to meet up with international regulatory developments whereas addressing native socio-economic realities. The transition from the present open regulatory setting to a sector-specific framework will probably be complicated, nevertheless it additionally provides a possibility for companies to construct better belief with their prospects via proactive engagement, because the method in direction of governing AI in South Africa is mapped.
* * * * *
Despina Lazanakis, Trainee Solicitor, has contributed to this authorized replace.
