South African organisations have significant gaps in cloud security, data governance and access controls as AI changes the cyber threat landscape, according to Accenture. The figures point to a practical problem: many firms are adding AI to foundations they have not yet secured.
Accenture AI Watch analysis
What happened
Accenture’s South Africa managing director, Kgomotso Lebele, sets out the concerns in a Consultancy.co.za article published on 2 November. The article cites Accenture research finding that only 18% of South African organisations have full visibility into cloud security configurations, while 6% have approved security baselines.
The research also found that just 18% have classified data by sensitivity, 24% have clear policies governing AI use, and 8% integrate security testing throughout software development. Accenture says AI can increase the speed and scale of some cyber tasks, including vulnerability discovery. These are figures and assessments attributed to Accenture, not an independent audit of every organisation in the country.
Why it matters
The gaps are not abstract future worries. AI projects depend on cloud services, access to data and decisions about who or what can use it. Weak controls in those areas can leave organisations with more exposure as their technology becomes more connected.
The figures also make a useful distinction: preparing for AI-related risks is not just a matter of buying an AI security tool. It includes knowing where sensitive data lives, limiting access and building security checks into software development.
Our read
Accenture’s warning is most useful where it gets specific. If the reported figures are representative, the unglamorous work of inventory, access control and testing deserves attention before the next AI pilot gets its own launch party. The numbers are a prompt for scrutiny, not a substitute for organisation-by-organisation evidence.
What to watch
- Whether Accenture publishes the research methodology and sample behind its South African figures.
- Whether South African organisations disclose measurable progress on cloud visibility, data classification and least-privilege access.
- Whether AI security tools are adopted alongside, rather than instead of, basic security controls.
Discussion spark: Should organisations have to meet minimum data, identity and cloud-security standards before deploying AI at scale, or should that remain a risk each business manages for itself?
Sources and evidence
- AI is raising South Africa’s cybersecurity stakes – Consultancy.co.za (2 November 2026, 20:03 UTC)
not affiliated with or endorsed by Accenture AI