Artificial intelligence has become an integral part of research and development across industries ranging from pharmaceuticals and biotechnology to mining, engineering, software development and advanced manufacturing. Businesses are increasingly using AI to analyse data, identify compounds, optimise processes, design products and accelerate innovation.

As AI becomes more deeply embedded in the innovation process, important questions arise. Who is the inventor when AI contributes to an invention? Who owns the resulting intellectual property? Can an AI-assisted invention be patented?

For businesses investing in innovation, these questions are becoming increasingly important.

The use of AI is not, in itself, a barrier to patent protection. The real issue is whether the human contribution to the invention can be identified, documented and supported.

AI as Part of the Innovation Process

Patent systems were developed around human inventors. While AI can now generate technical solutions, propose alternative designs or identify patterns that may not be readily apparent to researchers, the patent system continues to focus on the contribution made by people.

The distinction is an important one. A researcher who identifies the technical problem, determines the research objectives, selects the parameters for the AI system, evaluates alternative outputs, rejects unsuitable solutions and ultimately recognises the inventive concept occupies a very different position from someone who merely enters a prompt into an AI tool and accepts the first response.

The facts surrounding the development process therefore matter.

Businesses should view AI as a powerful research tool rather than a substitute for human ingenuity. A successful patent application should be capable of telling the story of the invention - from the technical problem that was identified, through the role played by AI, to the human insight that ultimately resulted in the inventive concept.

Inventorship and Ownership

Inventorship and ownership are closely related, but they are distinct legal concepts.

Inventorship concerns the individual or individuals who contributed to the inventive concept. Ownership concerns the person, company, university or research institution that is legally entitled to apply for and own the patent.

In commercial environments, ownership may arise through employment agreements, consultancy arrangements, collaboration agreements, research contracts, funding conditions or formal assignments.

Where AI tools are used during research and development, the relationship between inventorship and ownership can become more complex. A company may own the AI platform, fund the research and employ the development team, but it should still be able to identify the human inventors and demonstrate how the right to apply for patent protection has passed from those inventors to the patent applicant.

Businesses should also recognise that members of a research team may contribute in different ways. Some individuals may simply operate AI software, while others identify the technical problem, interpret AI-generated outputs, design experiments and recognise the inventive solution. These distinctions may become important when determining inventorship.

The South African Position

South African patent law does not yet contain a comprehensive statutory framework dealing specifically with AI-generated inventions. Nevertheless, South Africa attracted international attention in 2021 when it became the first country to grant a patent application naming an artificial intelligence system, known as DABUS (Device for the Autonomous Bootstrapping of Unified Sentience), as the inventor.

The DABUS application was filed by Dr Stephen Thaler, who argued that the invention had been autonomously generated by the AI system and that he, as the owner of the AI, was entitled to apply for and own the resulting patent rights. While several major patent offices, including those in the United States, the United Kingdom, Europe and Australia, rejected corresponding applications on the basis that an inventor must be a natural person, the South African application proceeded to grant.

The South African grant generated considerable international attention, with some commentators suggesting that South Africa had recognised AI systems as inventors. That conclusion should, however, be approached with caution.

Unlike many jurisdictions, South Africa currently operates a patent registration system rather than a system of substantive patent examination. Patent applications are generally examined only for compliance with formal requirements before grant and are not assessed by the Patent Office for novelty, inventive step, inventorship or entitlement to apply. Consequently, the grant of the DABUS patent did not involve a judicial or administrative determination that an AI system may legally qualify as an inventor under South African law.

Importantly, the DABUS patent has not been tested before the South African courts. Until there is judicial guidance or legislative reform, businesses should avoid assuming that the DABUS grant establishes a general legal principle. From a practical perspective, organisations should continue to identify and document the human contribution to AI-assisted inventions, ensure that inventorship can be properly supported, and confirm that ownership has passed to the patent applicant through the appropriate contractual arrangements.

Patent Strategy for AI-Assisted Innovation

As AI becomes increasingly sophisticated, questions surrounding inventorship, ownership and entitlement are likely to become more prominent and businesses should be able to reconstruct the development pathway that led to an invention. This means documenting not only the final technical solution, but also the decisions that were made throughout the research process.

Clear records may prove invaluable when preparing patent applications, identifying inventors, establishing ownership, responding to due diligence enquiries, negotiating licences or resolving future ownership disputes.

Ultimately, AI may assist in generating ideas, but successful patent protection will continue to depend upon identifying the human ingenuity that transformed those ideas into a patentable invention.