This publication refers to legislation, regulatory developments or market practice outside South Africa. Such references are provided for general information and comparative context only. Drawing on international developments to inform South African and cross-border commercial strategy is an important part of our advisory approach. Deneys is a South African law firm and does not practise foreign law. If you require advice on legal matters in another jurisdiction, we would be pleased to assist in coordinating appropriate support through trusted international relationships.
Rapid advances in artificial intelligence have shifted AI from a future concept into a core component of modern business strategy. Organisations are adopting AI and generative AI tools to improve efficiency and scale content creation, while global investment in AI infrastructure continues to grow. These developments have sharpened focus on how intellectual property laws interact with AI-driven innovation.
IP protection in the age of AI
IP regimes are built around recognising and rewarding human creativity. Patent and copyright systems generally expect identifiable human contribution. This is creating tension as AI systems produce increasingly sophisticated output.
Most jurisdictions continue to treat humans as the only permissible inventors in patent applications. Decisions across several countries, such as those examining Dr Stephen Thaler’s DABUS filings, have confirmed that AI systems cannot be listed as inventors because statutory language requires natural persons. Similar reasoning applies in copyright, where authorship is typically tied to a human creator and linked to terms of protection based on lifespan.
Where humans use AI as a tool, IP protection is generally more attainable. AI-assisted work still reflects human creativity and decision-making. The more difficult question arises when content is predominantly AI-generated. Different jurisdictions have adopted varying approaches, with some denying protection outright and others recognising the human’s selection, prompting or curation as sufficient involvement.
Ownership of AI-generated content
Ownership remains one of the most debated issues. While AI systems cannot hold rights, potential claimants include system developers, owners, trainers and end users. Contract terms increasingly play a determinative role, with some AI platforms assigning rights in output directly to users.
A few jurisdictions have experimented with sui generis rights for computer-generated works, but approaches vary widely. Businesses deploying AI at scale may benefit from reviewing contractual IP frameworks, licensing terms and internal policies to ensure clarity on ownership and exploitation rights.
Infringement and liability risks
AI introduces material risk exposure throughout the technology lifecycle. Litigation abroad has focused on whether training data infringes copyright where datasets contain protected material obtained without consent. Even where certain countries recognise limited exceptions for data mining, global training practices make infringement risk difficult to eliminate.
Output-related infringement is another concern. Generative models may unintentionally reproduce protected works, particularly in niche or data-sparse scenarios. Potentially liable stakeholders include developers, operators, corporate deployers and end users, depending on the factual context.
Mitigating commercial and legal risk
Organisations using AI for content creation or internal automation may wish to consider:
- implementing technical controls to reduce the chance of outputs replicating training materials
- using plagiarism or similarity tools before publishing AI-generated content
- training employees on safe prompting practices and avoiding requests that may solicit third-party material
- seeking contractual warranties or indemnities from AI solution providers, particularly for high-value use cases such as marketing assets, product documentation or customer-facing materials
Conclusion
AI continues to reshape the IP landscape, creating significant opportunities but also heightened uncertainty. Legal frameworks are evolving unevenly, and courts in multiple jurisdictions are addressing foundational questions for the first time. Businesses that rely on AI systems for product development, content creation or data processing may benefit from staying attuned to global developments and adopting risk-based governance strategies as part of broader innovation planning.
Latest news
