South Africa Directors Duties: Urgent Need for AI Regulation

• While South African company law does not yet have a standalone legislative framework specifically governing AI use by boards, the King V Code on Corporate Governance, effective from January 1, 2026, explicitly places responsibility on boards for the ethical and strategic oversight of AI.
• Directors are increasingly using AI for strategic decisions, including financial assessment, market trend prediction, and risk identification.
• While delegation to assisted or augmented AI may be permissible with proper oversight, delegating to autonomous AI is deemed inappropriate due to directors' non-transferable responsibilities.
• Concerns exist regarding AI's potential for bias from historical data, its 'black box' nature, and its inability to exercise human judgment or empathy.
• There is an urgent call for clear legal rules to address this regulatory vacuum and guide directors on their duties concerning AI-driven decisions.

Delegation to autonomous AI should be prohibited, as it would transfer decision-making power and absolve directors of their duties.

Artificial intelligence is increasingly influencing how boards of directors in South African companies make critical decisions. Directors are leveraging AI tools for various strategic functions, including assessing complex financial information, forecasting market trends, and proactively identifying potential risks before finalizing strategic choices. This integration marks a significant shift in corporate governance practices across the nation.

However, this rapid adoption highlights a critical regulatory void: While South African company law does not yet have a standalone legislative framework specifically governing AI use by boards, the King V Code on Corporate Governance, effective from January 1, 2026, explicitly places responsibility on boards for the ethical and strategic oversight of AI. Globally, many jurisdictions have already moved to address this, with the European Union, South Korea, and Vietnam implementing comprehensive AI legislation, while countries like China have introduced more targeted regulations.

A recent academic paper, focusing on company law implications, underscored that existing South African legal provisions provide minimal guidance on these emerging issues. As AI becomes an indispensable part of daily business operations, this regulatory gap is escalating into a significant legal and governance challenge. The research emphasizes the urgent need for clear legal rules to prevent a deepening legal vacuum within South African boardrooms.

Boards can deploy artificial intelligence in various capacities to manage companies and inform decision-making. The simplest form, known as assisted AI, handles administrative tasks, allowing directors to maintain full control over strategic choices. Examples include compiling meeting agendas, scheduling meetings, and drafting reports, thereby freeing directors to concentrate on higher-level strategic matters. A more advanced application is augmented AI, which excels at analyzing vast datasets, identifying patterns, and predicting potential outcomes for different business scenarios. For instance, it could evaluate merger risks and recommend optimal courses of action, with the board ultimately retaining the authority for the final decision.

The most sophisticated iteration is autonomous AI, capable of making decisions and executing tasks without continuous human intervention. Boards might, for example, authorize an autonomous AI system to make investment decisions within predefined parameters, independently monitor markets, and execute trades. While these AI applications offer clear advantages in supporting corporate decision-making, research also identifies significant limitations that can compromise the reliability of AI-generated recommendations.

One primary concern is that AI systems learn from existing data; if this data is inaccurate or biased, the AI's recommendations will likely reflect those biases. This is particularly pertinent in South Africa, where historical inequalities are often embedded in many datasets. A 2024 study by the Centre of Excellence in Financial Services, for instance, indicated that AI used in South African banking could inadvertently perpetuate racial bias. Without adequate oversight, reliance on such systems risks repeating or exacerbating existing discriminatory patterns. Another challenge arises from the 'black box' nature of many AI systems, where even their developers struggle to explain how a particular conclusion was reached. Furthermore, AI inherently lacks essential human qualities for effective board leadership, such as empathy, intuition, and emotional intelligence.

Under South African company law, directors are permitted to delegate certain functions to other individuals or entities. However, this delegation does not absolve them of their ultimate responsibility for the company's management; they remain fully accountable for all board decisions. The principle, according to recent research, should extend to the use of AI. Directors may appropriately delegate functions to assisted AI and potentially to augmented AI, provided they maintain rigorous supervision over the AI system and do not relinquish their fundamental responsibilities.

Crucially, delegation to autonomous AI should be prohibited. Allowing autonomous AI to make decisions would effectively transfer the decision-making power from directors, thereby enabling them to abandon their statutory responsibilities. Directors are legally bound to exercise independent judgment and to act at all times in the company's best interests; they cannot simply accept AI recommendations without critical evaluation.

The 'black box' problem further complicates this duty. If directors cannot comprehend the reasoning behind an AI's recommendation, they are fundamentally unable to determine whether that recommendation genuinely serves the company's best interests. This lack of transparency directly conflicts with their legal obligations for independent judgment and fiduciary duty.

The current legal vacuum surrounding South Africa directors duties AI regulation poses a growing challenge for corporate governance AI South Africa. As AI becomes increasingly integrated into the fabric of business operations, the absence of clear legal rules creates significant uncertainty regarding directors' responsibilities AI and the permissible scope of delegation to AI South Africa.

There is an urgent and pressing need for specific legislative guidance to clarify these issues. Without it, boards face increasing legal and governance risks, particularly concerning accountability for AI-driven decisions. Establishing clear parameters for the use of assisted, augmented, and autonomous AI is essential to ensure that directors can leverage technological advancements while upholding their statutory duties and mitigating potential liabilities.

Lawyers and compliance officers must advise South African boards on navigating the current regulatory vacuum regarding AI use, particularly concerning directors' duties, delegation limits, and accountability for AI-driven decisions, given the lack of specific company law guidance. They should also monitor for impending legislation to mitigate legal and governance risks.

Source: Insights derived from recent company law research.