Melissa Cogger, Partner at Bowmans

Artificial intelligence (AI) is rapidly transforming the world of work. From automated CV screening and chatbot-led interviews to algorithmic performance monitoring and predictive workforce analytics, employers are increasingly deploying AI systems across the employment lifecycle. These tools promise efficiency gains and data-driven objectivity. However, they also present significant legal risks, particularly in the areas of privacy, non-discrimination and fair labour practices.

This article provides an overview of the South African regulatory landscape as it pertains to AI in employment and recruitment.

The South African National AI Policy Framework

In August 2024, the South African Government published the National Artificial Intelligence Policy Framework (AI Policy Framework). While the AI Policy Framework does not constitute binding legislation, it represents the Government’s strategic vision for the governance of AI in South Africa and signals the direction of future regulation. The AI Policy Framework is built upon several strategic pillars, namely:

• Ethical AI guideline developments – addressing bias, fairness, transparency and accountability in AI systems;
• Privacy and data protection – ensuring AI systems comply with existing data protection legislation, principally the Protection of Personal Information Act, 4 of 2013 (POPIA);
• Safety and security – implementing cyber security protocols and risk management frameworks;
• Transparency and explainability – requiring that AI-driven decisions can be understood and interrogated by affected persons;
• Fairness and mitigating bias – proactively identifying and addressing algorithmic bias;
• Human control of technology – maintaining a human-centred approach to AI deployment; and
• Professional responsibility – ensuring that professionals deploying AI remain accountable for outcomes.

The AI Policy Framework should be distinguished from the Draft South Africa National Artificial Intelligence Policy that was published for public comment in April 2026. That draft policy document was withdrawn shortly after it was gazetted as a result of AI hallucinations citing information that did not exist.

The current legislative framework

South Africa does not yet have a dedicated AI Act. The regulation of AI in the employment context is accordingly governed by existing legislation of general application. Of particular relevance to human resources (HR) is POPIA (which regulates the processing of personal information, including through automated means), the Employment Equity Act, 1998 (EEA) (which prohibits unfair discrimination in the workplace), the Labour Relations Act, 1996 (LRA) (which protects employees against unfair dismissal and unfair labour practices), and the Regulation of Interception of Communications and Provision of Communication-Related Information Act 70 of 2002 (RICA) (which regulates the interception and monitoring of certain communication-related information), to name a few. These Acts were not drafted with AI in mind, but their provisions are sufficiently broad to capture many of the risks associated with AI deployment in HR.

The ‘human in the loop’ principle

A central theme of the AI Policy Framework, and of international best practice more broadly, is the principle of keeping the ’human in the loop’. This principle requires that AI systems should not operate in isolation when making decisions that have significant consequences for individuals.

In the HR context, this means that AI should serve as a tool to assist human decision-makers, not to replace them. Every material employment decision, whether it relates to hiring, performance assessment, promotion or dismissal should ultimately be made or ratified by a human being who can exercise judgement, apply context and be held accountable.

AI in recruitment and employment: practical examples

The following are examples of how AI is currently reshaping HR practices:

• automated screening of curricula vitae and application forms, using natural language processing to rank candidates;
• chatbot-led initial interviews and psychometric assessments;
• predictive analytics to identify candidates most likely to succeed in a role or most likely to resign;
• algorithmic performance monitoring and productivity scoring;
• AI-driven pay benchmarking and remuneration recommendations; and
• automated identification of employees for retrenchment based on selection criteria applied by algorithms.

Each of these applications gives rise to distinct legal considerations and risks under POPIA, the EEA and the LRA, which will be explored in the subsequent parts of this series.

Conclusion and next steps

While South Africa’s regulatory framework for AI remains in its nascent stages, the existing legislative architecture provides meaningful safeguards. Employers who deploy AI in HR must navigate a complex web of obligations under various legislation.

ENDS