South Africa’s insurance industry is entering a new phase of digital transformation as providers increasingly explore artificial intelligence (AI) to improve operational efficiency, strengthen risk management and enhance customer service. However, the absence of a finalised national AI policy means insurers must take greater responsibility for establishing internal rules governing how the technology is deployed.
The issue gained renewed attention following the release of KPMG South Africa’s 2026 insurance industry survey, which examines the opportunities and risks shaping the local insurance market. The research places AI alongside cybersecurity, changing customer expectations, geopolitical uncertainty and evolving business risks.
For insurers, the challenge is no longer simply whether to adopt AI. It is how to use it responsibly while protecting sensitive customer information and maintaining confidence in decisions that can affect people’s financial security.
A national AI policy still in development
South Africa’s progress towards a national AI framework encountered a setback in April 2026, when the Department of Communications and Digital Technologies withdrew its draft National Artificial Intelligence Policy after fictitious references were identified in the document.
The withdrawal followed concerns about the reliability of the research underpinning the proposed framework. The incident also highlighted a central lesson for organisations adopting AI: technology-assisted work still requires robust verification, professional judgement and accountability.
The government subsequently indicated that it would rework the policy, with the aim of establishing appropriate national standards for ethical AI use. The policy-development process remains important to businesses seeking clearer guidance on responsible implementation.
In the meantime, the lack of a final national AI policy does not mean that South African companies operate without legal obligations. Existing requirements, including the Protection of Personal Information Act (POPIA), remain relevant when organisations collect, analyse, store or use personal information through automated systems.
For insurers, this creates a need to align AI initiatives with existing legislation while preparing for a more developed national governance framework.
How insurers are beginning to use AI
AI can support several insurance activities, particularly those involving large volumes of information and repetitive administrative work.
Examples include assisting employees with documents and correspondence, helping customer-service teams respond to routine enquiries, identifying unusual claims patterns and supporting preliminary underwriting assessments.
Claims processing is another area where automation may improve efficiency. AI tools can help organise submitted documents, extract relevant information and flag cases that require additional investigation. Fraud-detection systems can also identify patterns that may warrant closer examination by experienced claims professionals.
However, these applications should not be confused with fully autonomous insurance operations. The use of AI to assist with a task does not necessarily mean that a system independently makes the final decision.
KPMG’s 2026 survey discusses the importance of understanding the relationship between technological ambition, business value, risk management and human judgement. This is particularly relevant in insurance, where an incorrect decision can affect whether a customer receives compensation, obtains cover or faces different pricing.
The immediate opportunity is therefore to use AI to support employees and improve processes without removing appropriate human oversight.
Why insurers need their own AI frameworks
Without a final national AI policy, insurers need internal governance arrangements that define acceptable uses of AI, establish accountability and identify risks before systems are deployed.
A practical framework should begin with an inventory of AI applications across the business. Organisations need to understand which tools employees are using, what information those systems process and whether sensitive customer data could be exposed to external platforms.
Clear approval procedures are equally important. Employees should know which AI tools are authorised, what information may be entered into them and when a human reviewer must verify the output.
Insurers should also test systems for accuracy, reliability and potential bias. If historical information contains unfair patterns or incomplete data, AI-assisted assessments may reproduce those weaknesses.
Governance must extend to external technology providers. Insurers relying on cloud services, software vendors and AI platforms need to understand how customer information is handled, where it is stored and what safeguards apply if a supplier experiences a security incident.
These measures can help companies establish consistent controls even while national policy development continues.
Data privacy and cybersecurity move to the forefront
Insurance companies process highly sensitive information, including identity details, financial records, vehicle information and, in some products, medical data. This makes responsible data handling particularly important when AI systems are introduced.
POPIA establishes obligations concerning the lawful processing and protection of personal information. Section 71 also addresses certain decisions based solely on automated processing that may have legal consequences or significantly affect individuals, subject to the provisions and exceptions in the Act.
Insurers should therefore assess whether their AI applications could trigger these requirements and ensure that appropriate safeguards, explanations and review procedures are in place.
Cybersecurity adds another layer of complexity. AI can strengthen threat detection and help teams investigate suspicious activity, but criminals can also use AI to create convincing phishing messages, impersonate customers or employees and automate aspects of cyberattacks.
Third-party exposure is another concern. A vulnerability at a technology supplier can affect an insurer even if its own systems remain secure. Effective governance must therefore consider the broader network of service providers and connected platforms supporting insurance operations.
Balancing innovation with customer trust
AI adoption will be most valuable when measurable improvements are accompanied by clear accountability.
Insurers should evaluate whether a system genuinely reduces processing times, improves accuracy or strengthens customer service. They should also monitor errors, complaints, unexpected outcomes and the costs of maintaining the technology.
Human oversight remains especially important for high-impact decisions. AI-generated recommendations should be reviewable, and staff must have the authority and knowledge to challenge outputs that appear inaccurate or unfair.
Customers should not be left uncertain about how their information is used or how to query a decision affecting their cover or claim. Clear communication and accessible complaint channels can help maintain trust as automated processes become more common.
What comes next for insurance technology?
The next stage of AI adoption in South Africa will depend on the ability of insurers to move beyond experimentation and establish reliable, secure and commercially valuable systems.
A revised national AI policy could provide greater consistency, but companies do not need to postpone responsible governance while waiting for it. Internal controls, staff training, independent testing and stronger supplier oversight can already help reduce risks.
For technology leaders, risk committees and insurance executives, the priority is to treat AI governance as a business responsibility rather than a technical afterthought.
The bottom line: South African insurers have an opportunity to improve efficiency and develop more responsive services through AI. Their long-term success, however, will depend on protecting customer information, retaining meaningful human oversight and demonstrating that innovation can be trusted.
Read More: https://kpmg.com/za/en/insights/2026/10/insurance-industry-survey-2026
