KAVI PATHER: AI adoption racing ahead in SA without guardrails

In the boardrooms of SA’s largest corporations, AI is no longer a futuristic buzzword — it’s a present-day reality. 

Our recent round-table survey with many of SA’s largest companies, revealed that nearly all are either experimenting with or fully implementing AI technologies. 

From integrating Microsoft’s CoPilot for enhanced productivity to deploying AI in call centres for smarter customer interactions, the enthusiasm is palpable. These tools promise to revolutionise operations, boost efficiency and drive innovation in an economy hungry for growth. 

Yet beneath this wave of adoption lies a stark vulnerability: a glaring absence of enterprise-wide AI risk and governance frameworks. The survey paints a concerning picture. While AI permeates customer-facing applications and internal processes, few organisations have a comprehensive approach to managing its risks. 

Alarmingly, most lack even a basic inventory of where AI is being used within their operations. This blind spot means companies are unwittingly exposing themselves to unidentified and unmanaged risks ranging from data privacy breaches and biased decision-making to operational failures that could erode trust and invite reputational damage. 

In a country such as SA, where economic inequality and regulatory scrutiny are already high, such oversights could amplify existing challenges, potentially widening social divides or triggering unforeseen legal liabilities. The urgency is clear: these risks demand immediate attention. 

SA may not yet have dedicated AI regulations, but that shouldn’t be an excuse for complacency. In the interim, businesses would be wise to benchmark against established international standards, such as the National Institute of Standards & Technology AI Risk Management Framework, the Institute of Electrical & Electronics Engineers’ ethics guidelines, or the EU’s AI Act. 

These provide robust blueprints for governance, helping companies navigate the ethical and operational minefields of AI deployment. By proactively adopting such assessments, SA firms cannot only mitigate risks but also position themselves as responsible leaders in a global AI landscape. This call for action aligns with a broader global conversation on AI assurance. 

As EY, in collaboration with the Association of Chartered Certified Accountants, has noted in recent analyses, well-designed AI assessments are pivotal in evaluating whether investments in the technology align with business and societal expectations. 

Around the world leaders, policymakers and investors are grappling with AI’s dual nature: a catalyst for productivity and innovation, yet often perceived as an opaque “black box” lacking transparency, reliability and trustworthiness.

The EY 2025 AI Sentiment Index Study surveying more than 15,000 people globally underscore this tension. While 82% of consumers have embraced AI in recent months, 58% worry that organisations are not holding themselves accountable for its negative effects. 

In SA, where AI could transform sectors such as mining, finance and agriculture — potentially adding billions to GDP — these concerns resonate deeply. Business leaders here are asking critical questions: how can we ensure AI systems are safe and effective? How do we identify and manage risks? How do we measure performance against governance criteria?

Understanding the emerging landscape of AI assessments is key. Policymakers in nearly 70 countries had introduced more than 1,000 AI-related initiatives by January 2025, many incorporating assessment frameworks often dubbed “AI audits” or “AI assurance.” These can be voluntary or regulatory, helping businesses evaluate performance, governance and compliance. Broadly, they fall into three categories:

• Governance assessments scrutinise whether appropriate internal policies, processes and personnel are in place to oversee AI systems. This includes risk management, suitability and reliability — directly addressing the survey’s finding that few SA companies track AI usage enterprise-wide.
• Conformity assessments verify compliance with laws, regulations, standards or policies. For SA firms eyeing international markets, aligning with frameworks such as the EU AI Act could prevent trade barriers and enhance competitiveness.
• Performance assessments quantify how well AI functions, using metrics for accuracy, non-discrimination and reliability. In customer-facing applications such as AI-driven call centres, these ensure fairness and efficacy, which are crucial in a diverse society such as ours.

However, not all assessments are created equal; variations in quality can undermine their value. To maximise effectiveness assessments should embody certain core characteristics. Specificity is paramount: clearly define the objective, scope and subject matter to avoid vague outcomes. A clear methodology ensures consistency, with defined criteria and terminology allowing for comparable results — whether the assessment yields explicit conclusions or procedural summaries.

Equally vital are the qualifications of assessors. Providers must demonstrate competence, objectivity and professional accountability to lend credibility. In SA, where local expertise is growing but still nascent, partnering with accredited bodies or international firms could bridge gaps.

For SA business leaders, the path forward involves strategic next steps:

• Integrate AI assessments into corporate governance and risk management. They can illuminate evolving risks and confirm if systems deliver as promised, turning potential liabilities into strengths.
• Even without regulatory mandates, consider voluntary assessments to foster confidence among stakeholders. Market pressures, investor expectations or internal ethics may demand it. For instance, in industries such as banking or healthcare, voluntary audits could differentiate ethical players, attracting talent and capital.
• Select the right assessment type — governance, conformity or performance — and decide on internal versus third-party execution. Third-party involvement often adds impartiality, especially for high-stakes AI applications.

SA’s AI surge is a golden opportunity, but unchecked enthusiasm risks derailment. By embracing well-designed assessments, companies can build trustworthy systems that drive sustainable growth. Policymakers, too, should accelerate local frameworks, drawing from global best practices. 

Ultimately, proactive governance isn’t just prudent — it’s essential for harnessing AI’s promise while safeguarding our future. As we stand at this technological crossroads, let’s ensure SA businesses lead with foresight, not hindsight.

• Pather is EY Africa AI leader.