Artificial Intelligence Lawyers | south africa

AI law provides crucial guidance and protection in the rapidly evolving field of artificial intelligence. As AI becomes more pervasive, clients require legal expertise to address ethical, regulatory, and liability challenges, ensuring they can harness AI’s potential while mitigating risks and complying with relevant laws. Failure to understand and adhere to AI-related legal considerations can lead to costly legal disputes, reputational damage, and potential business disruptions.

The development and use of AI entail important legal considerations, including considerations pertaining to  intellectual property rights, data privacy and security, liability and accountability, transparency and explainability, ethical implications, regulatory compliance, contractual arrangements, international legal frameworks. Addressing these legal aspects is essential to ensure responsible and compliant utilization of AI technology, as failure to do so can lead to significant risks, liabilities, and potential harm to individuals and society.

Currently, there is no formal legislative position for AI that has been implemented. However, the Artificial Intelligence Act promulgated by the EU has reached an overwhelming majority vote as it begins its entry into formal legislation. In addition, many countries around the world are working at speed attempting to put in place a regulatory regime that is sustainable to keep up with this vibrant technology. International Bodies such as the the Organisation for Economic Co-operation and Development (OECD) and the World Economic Forum (WEF) have also issued principles and guidelines on AI ethics and governance which are critical for legal frameworks in organisations aiming to be ahead of the curb.  ness, reliability, location, health, personal preferences or conduct.”

AI ethics should take into account privacy and security, reliability and safety, fairness and inclusiveness, transparency and accountability.

South Africa does not yet have a single, dedicated law regulating artificial intelligence. AI use is currently governed through a patchwork of existing legislation – POPIA where personal information is processed, consumer protection law where AI affects consumers, sector-specific regulation where it applies (financial services, for example), and general principles of contract, delict, and IP law. Government has published policy positions signalling the direction of future regulation, but businesses currently need to build AI governance around the existing legal framework rather than a single, dedicated “AI Act.”

This is genuinely unsettled under South African law. The Copyright Act currently defines authorship in terms of a human (or, for certain computer-generated works, “the person by whom the arrangements necessary for the creation of the work were undertaken”), which was not drafted with generative AI in mind. In practice, most businesses address this contractually – through the terms of service of the AI tool itself (which frequently assign usage rights, though not always full ownership, to the user) and through their own agreements with staff, contractors, and clients about how AI-assisted output is owned and used. Until the position is clarified by legislation or the courts, contractual clarity is the safest protection.

An effective AI use policy should set out which AI tools are approved for use in the business, what categories of information may (and may not) be input into them – particularly personal information, client data, and confidential or privileged material – clear rules on reviewing and verifying AI-generated output before it’s relied on or sent externally, disclosure expectations (internally and to clients, where relevant), and accountability for the final work product. It should also address which tools are prohibited, and how new tools get approved. A policy nobody has read isn’t a policy – it needs to be built into onboarding and refreshed as the tools your business uses change.

Key terms to negotiate in an AI vendor contract include clear ownership of your input data and any output generated using it, restrictions on the vendor using your data to train its models (or clear opt-out rights if it does), indemnities covering IP infringement and inaccurate or harmful output, meaningful service levels and liability caps proportionate to how the tool will be used, and advance notice of material changes to the underlying model or its behaviour. AI vendor terms are often drafted heavily in the vendor’s favour and change quickly – treating them as a standard SaaS agreement without this additional scrutiny is one of the most common gaps we see.

Yes. If an AI tool processes personal information – whether that’s customer data, employee data, or information contained in documents fed into the tool – POPIA applies in the same way it would to any other form of processing. This means you need a lawful basis for the processing, appropriate safeguards, and clarity on where the data goes, including whether it’s used to train the underlying model and whether it’s transferred outside South Africa. Many popular AI tools are hosted and trained internationally, which makes the cross-border transfer and data retention questions particularly important to work through before an AI tool is rolled out across a business.

AI can be used to support recruitment and screening, but it needs to be deployed carefully. POPIA restricts decisions based solely on automated processing (including profiling) that significantly affects a data subject, unless specific conditions are met — so an AI screening tool generally can’t be the sole and final decision-maker without appropriate safeguards, such as human review. There’s also a discrimination risk to manage: if an AI tool’s screening criteria produce a disparate impact on any protected group, this can expose an employer to an unfair discrimination claim under the Employment Equity Act, regardless of intent. Businesses using AI in recruitment should keep a documented, human decision-maker in the loop, and periodically test the tool’s outcomes for bias.

Caveat Legal’s AI and technology lawyers advise South African businesses on AI governance, data protection, technology contracts, intellectual property and emerging regulatory risk. Caveat combines experienced legal advice with a practical understanding of how AI is being adopted in business.