This is the final article in our series on AI and intellectual property for South African financial services firms. Having addressed copyright infringement, patents and trade secrets, this article draws the threads together into a practical, board-level agenda.
AI is no longer an information technology issue. For South African financial services firms, it is a governance issue, and the board needs to take ownership of it.
IP ownership is a strategic asset question
If your firm generates valuable outputs using AI, including credit models, client analyses, regulatory reports or trading algorithms, does IT own them? Without express contractual provisions, South African copyright law may provide an outcome that is not commercially appropriate, and the position on AI-generated works remains unresolved. The board should ensure that IP ownership in AI-generated work is addressed in every material vendor, development and client-facing contract.
Your training data exposure may become your risk
If the AI tools your firm procures were built on unlicensed copyright material, your firm may face indemnity gaps if a claim is brought against the provider. The Getty appeal in the UK and ongoing US litigation mean that the risk profile of commercially deployed AI systems is still evolving.
The board should consider whether procurement contracts provide adequate indemnity protection and whether those indemnities are supported by sufficiently robust providers.
Confidential information is your most exposed asset
The risk that proprietary information, including client data, trading strategies and model methodologies, enters a commercial AI system and is lost to the firm permanently is real and often underestimated.
Once confidential information has been input into an external system, it may be used to train that system's models and may not be recoverable. The board should ensure that governance policies, access controls and contractual restrictions are in place and consistently applied.
Your AI contracts may not protect you
Procurement decisions made at an operational level may result in contracts that are silent on training data provenance, output ownership, IP indemnities and POPIA compliance. A structured contract review is a practical and urgent priority, not a legal formality.
Patent rights may be slipping away unnoticed
If your firm is developing novel AI-driven processes in areas such as fraud detection, credit scoring, algorithmic trading or regulatory reporting, the window for patent protection may be closing.
Public disclosure of a product or process can extinguish patent rights before an application is filed. The board should ensure there is a process in place to identify and protect patentable innovation before it enters the public domain.
The regulatory horizon
Three jurisdictions warrant close monitoring:
-
United Kingdom: The House of Lords Communications and Digital Committee published a report in March 2026 recommending greater transparency in AI training data and development of a fair licensing market. The Getty appeal is pending, and the UK position is likely to evolve in the next 12 to 18 months.
-
European Union: Most provisions of the AI Act apply from August 2026. Obligations on general-purpose AI model providers, including training data transparency and copyright compliance policies, are already in force. Firms with an EU nexus, or that procure from EU-regulated providers, should understand how these obligations apply contractually.
-
United States: The fair use position remains unresolved in the courts. Given that most AI systems are trained in the US, the outcome of ongoing litigation will directly affect the risk profile of tools used by South African firms.
Domestically, developments such as the Copyright Amendment Bill, the FSCA's Innovation Hub and the SARB's AI governance initiatives should also be monitored.
Immediate board priorities
- Appoint an AI governance lead responsible for monitoring legal and regulatory developments and escalating material issues to the board;
- conduct an IP audit of all AI tools currently in use, whether procured or developed in-house, to identify gaps in ownership, licensing and protection;
- review all AI procurement contracts for IP ownership provisions, training data indemnities, confidentiality protections and data processing terms;
- establish a trade secret register and implement appropriate protective measures;
- train employees on what information must be kept confidential and why; and
- build a regulatory monitoring programme that provides regular updates to the board.
The legal framework governing AI and intellectual property continues to evolve. Firms that take a proactive approach, identifying their IP assets, protecting them appropriately, and building the internal capability to manage this changing landscape, will be better positioned than those that do not.
That is the central message of this series, and a key consideration for your board's AI agenda.
