Patentability of Pharmaceuticals and Diagnostics Using Artificial Intelligence in Australia
Artificial intelligence (AI) is transforming pharmaceutical and biotech research globally, and Australia is no exception. From accelerating drug discovery pipelines to providing new diagnostic tools for early disease detection, AI is helping researchers develop new therapeutics and medical intervention earlier, faster and more efficiently. However, this innovation brings with it critical questions about protection of intellectual property (IP) and developing IP management strategies.
The Role of AI in Drug Development and Diagnostics
Globally and locally, research institutes, large pharma and biotech start-ups are increasingly turning to AI to enhance drug discovery and diagnostic development by addressing inefficiencies in traditional methods which often result in high costs, long timelines, and low success rates.
AI tools are currently implemented in every stage of the pharmaceutical product life cycle, such as target protein identification and structure prediction, de novo drug selection, design and optimisation, along with toxicity/bioactivity/physicochemical property prediction and formulation development. [1], [2]
Likewise, AI is being implemented across a broad range of technical fields in the areas of diagnostic development. These include deep learning models for analysing histopathological patient samples, rapid ‘omics’ analysis aiding in the diagnosis of diseases and guiding personalised treatment strategies, as well as improving patient care with real-time monitoring and early risk detection.[3]
While the use of AI tools accelerates innovation, it also challenges the traditional framework for protecting the IP underpinning these inventions. Some key requirements of the Australian patent system impacted by the use of AI tools in the pharmaceutical and diagnostics innovation space are inventorship, inventive step (obviousness) and sufficient disclosure requirements.
Inventorship and Ownership: The Human Element
A central issue when considering AI under Australian patent law is inventorship, and by association through transfer of rights, ownership.
Under the Australian Patents Act 1990, only a ‘natural person’ can be listed as an inventor. This was clarified by the Full Federal Court decision in Commissioner of Patents v Thaler [2022] FCAFC 62, which held that an AI system (in this case, DABUS) cannot be an inventor under Australian patent law.[4]
IP Australia’s Patent Manual of Practice and Procedure explicitly states:
“artificial intelligence is not within the meaning of the term inventor and as such artificial intelligence cannot be not named as an inventor. A human(s) who has materially contributed to the final invention is considered an inventor…”.
Accordingly, any patent application must demonstrate that a human inventor was responsible for the conception of the invention, and that said human has made a qualitative contribution to the invention – even if AI tools were used during development.[5]
A further consideration for applicants when implementing AI tools in development processes is one of ownership. Applicants must be aware that IP rights may not subsist in any outputs generated by using AI tools in the first place and/or any IP rights may be transferred to the provider of the AI tool as a result of assignment.
Inventive Step (Obviousness)
Another key patentability requirement to be considered when implementing AI tools is that of inventive step.
Australian patent law requires that an invention must not be obvious to a person skilled in the art (PSA) in light of the common general knowledge (CGK) and/or the prior art available at the earliest priority date of the claims.
The test for assessing inventive step can be summarised as whether the PSA, informed by the relevant CGK and/or prior art, would be lead as a matter of course to try the claimed invention in the expectation it might provide a solution. [6]
As AI tools become more accessible and routinely implemented into everyday practice, it is foreseeable that patent offices may move to argue that AI-assisted discoveries introduce a broader CGK base due to ease of access to information of varied subject matter, thereby setting the requirements to establish an inventive step higher than would otherwise be expected.
Accordingly, patent applicants seeking to implement AI tools in the development of pharmaceuticals and diagnostics should avoid simply answering a question to a known problem using AI. Rather, a particular focus should be on the ‘unexpected or surprising’ outcome observed, or the improvement achieved over current methodologies when the invention is reduced to practice.
Sufficiency
A third aspect which must be considered by patent applicants implementing AI tools is the sufficiency requirement.
Australian patent law requires that a complete patent specification must disclose the invention in detail which is sufficiently clear and complete enough for it to be performed by a PSA.
As noted in IP Australia’s Patent Manual of Practice and Procedure:
“This provision reflects a fundamental principle of international patent law; in exchange for the exclusive rights given to the patentee, the patentee must share with the public the information necessary to make and use the invention.” [7]
As such, patent applicants utilising AI tools in drug development and diagnostics should considering providing sufficient detail to:
- The AI methodology employed, such that a PSA could achieve the claimed invention,
- The nature and source of data used by the AI-tool to achieve the claimed invention; and
- The role of human input, oversight or interpretation of the outputs and how these are reduced to practice.
Where disclosure may reveal commercially sensitive algorithms or datasets, applicants might consider a dual strategy of patent and trade secret protection for different aspects of their innovation pipeline. This approach will likely be driven by commercial considerations; however, applicants should be careful to strike the right balance between meeting the sufficiency requirements for obtaining patent protection and optimising their IP management strategy to adequately protect the unique and commercially valuable aspects of their invention.
Best Practices for Applicants in Australia and Looking forward
To strengthen patent applications involving AI-assisted pharmaceutical and diagnostic discovery, Australian innovators should:
- Identify and document the human inventors’ specific contributions, and contrast these with AI assisted contributions.
- Provide a detailed technical description of how AI tools were used and for what purpose.
- Disclose sufficient information about the AI tools and relevant data inputs that a PSA would need to reproduce the invention.
- Be aware of the terms and conditions associated with any AI tool, particularly freely available tools. It is vital to ensure the user understands IP ownership rights in the outputs generated by the AI tool, and when using these tools, applicants adjust their IP strategy accordingly.
With international debate growing over the role of AI in innovation, future challenges will no doubt be presented for innovators and IP professionals alike. Therefore, it is essential that applicants fully understand the implications of using AI-tools and how these may influence their IP strategies.
If you have questions about the topics of Pharmaceuticals, Diagnostics, AI tools and IP, feel free to contact the MBIP team for guidance. Our experienced attorneys keep up to date with the latest trends and developments in Australian IP law and would be happy to assist.
[1] Rumel Dey*, Dona Roy Chowdhury and Md Al Hafiz. Artificial Intelligence (AI)’S Transformative Impact on Drug Development and Regulatory Affairs: A 2025 Review. Am J Biomed Sci & Res. 2025 27(4) AJBSR.MS.ID.003572, DOI: 10.34297/AJBSR.2025.27.003572
[2] First AI-designed drug, Rentosertib, officially named by USAN, published 14 March 2025 – Drug Target Review.
[3] Takanobu Hirosawa*, Taro Shimizu. A Narrative Review of Artificial Intelligence in Medical Diagnostics. Computers, Materials & Continua 2025, 83(3), 3919-3944. https://doi.org/10.32604/cmc.2025.063803
[4] Commissioner of Patents v Thaler [2022] FCAFC 62
[5] IP Australia Patent Manual of Practice and Procedure 7.2.8.5, Relevant Law – Entitlement/Inventorship, AI-Assisted Inventions
[6] IP Australia Patent Manual of Practice and Procedure – 5.6.6.3 Tests for inventive step
[7] IP Australia Patent Manual of Practice and Procedure – 5.6.7.2 Clear enough and complete enough disclosure s40(2)(a)
