The United States Patent and Trademark Office’s (USPTO) Inventorship Guidance on AI-assisted inventions marks a crucial step in the evolving role of AI in the patent system, as outlined in a recent article co-authored by LML members. The article is published in the Journal of Intellectual Property Law & Practice, and co-authored by Prof. Kathy Liddell and Prof. Mateo Aboy, along with Assistant Professor Aparajita Lath of the National Law School of India University.
The growing use of artificial intelligence (AI) in innovation raises challenging questions of inventorship in patent law. This article analyses the USPTO’s 2024 Inventorship Guidance for AI-Assisted Inventions, which re-affirms a strictly human-centric approach to inventorship: only natural persons can be named as inventors, even if this means a clever invention generated principally by an AI system would be inventor-less and unpatentable. As the authors put it, “While AI can augment the inventive process, it cannot independently fulfil the role of an inventor. AI will be treated as an advanced tool”. This is “a cautious position”, the authors say, “with far-reaching practical, ethical, and legal implications as AI technology advances”.
The Guidance explains how to identify human contributions that are sufficient for inventorship in situations of AI-assisted innovation.Citing the US Federal Circuit Court’s decision in Pannu, the USPTO firmly states that only natural persons —i.e. human beings—who make a “significant contribution” to the conception of an invention can be recognised as inventors. Such contributions may include defining specific problems, modifying AI-generated outputs, or developing specialised systems. The Guidance also sets out a non-exhaustive list of ‘guiding principles’ to assist in evaluating inventorship in AI-assisted contexts.
The authors critically analyse the overarching requirement for a significant human contribution, the details within the USPTO’s guiding principles, and practical steps for improving compliance with the Guidance. They address key questions including: Was it appropriate to re-purpose the Pannu decision? Can one meaningfully distinguish between ‘specific’ and ‘general’ prompts, training and modifications? To be recognised as an inventor in an AI-assisted context, must the human contribution exceed what could be expected from a person with ordinary skill in the art? And is law reform desirable?
The authors argue that the current Guidance contains fundamental ambiguities likely to generate considerable legal uncertainty. Over time, the central principle that an inventor must be a human being will also result in a growing class of inventor-less inventions potentially chilling investment in AI-generated discoveries, particularly in areas such as drug development. So while the USPTO’s Inventorship Guidance on AI-assisted inventions marks a crucial step for the evolution of the patent system in our AI-era, the authors suggest it is likely to require revision and refinement.
The article is openly accessible here and forms part of a growing body of LML research at the intersection of AI, law, and biomedicine.