Demise of the machines: Full Court of the Federal Court of Australia overturns ruling on AI as a patent ‘inventor’
It was a short-lived reign for the machine in Australia.
After a somewhat surprising victory in the Federal Court last year1 (about which we wrote in our article ‘Rise of the Machines’), the hopes that an artificial intelligence machine, nicknamed DABUS, could be named as the inventor on a patent application have now been dashed by the Full Court2. In short, the Full Court upheld an appeal by the Commissioner, ultimately finding that Dr Thaler’s patent application had indeed lapsed for failure to name an ‘inventor’ as required by the relevant regulations3.
The five-judge bench of the Full Court handed down its 123-paragraph judgment just one month after hearing the appeal. It overturned the controversial decision of the primary judge, his Honour Justice Beach, in what can be regarded as a return to the orthodox application of principles of statutory construction.
The Full Court’s approach to statutory construction
In the absence of a statutory definition of the word “inventor” in the Patents Act 1990 (Cth) (the Act), the Full Court turned to a detailed consideration of the text of the legislation itself (“the surest guide to ascertaining the legislative intention”4) against the backdrop of its legislative history5. The Full Court paid particular regard to s 15 of the Act, which governs who may be granted a patent and in which the word “inventor” appears multiple times.
Taking this approach, the Full Court held that the intention of the Act (and subordinate regulations) was that the word “inventor” refer to a natural person only, specifically the person responsible for the inventive concept6. In the Full Court’s view, the primary judge’s broader interpretation was inconsistent with a natural reading of the language and purpose of both s 15 and the Act as a whole7.
The Full Court clarified that while there are multiple routes to patent entitlement provided for by s 15, all require there to be a legal relationship between the actual (human) inventor and the person first entitled to the grant, which is a legal impossibility in circumstances such as the present case where the purported inventor has no legal identity and therefore cannot give effect to an assignment8.
The Full Court appeared critical of the primary judge’s approach to statutory construction, by reference to what the Court itself might regard as desirable policy, imputing that policy to the legislation and then characterising that as the purpose of the legislation9, rather than the orthodox text-based approach taken by the Full Court10.
Conclusion
What is most interesting, perhaps, is the Full Court’s concluding remarks that the underlying policy questions thrown up by this litigation, which were (correctly) not the subject of considered debate or opinion, “should be attended to with some urgency”11. For example, whether “inventor” should be redefined to include AI and if so, to whom should a patent be granted in respect of its output? Further, would such an approach require a further recalibration of the standard of inventiveness?12
So, while Dr Thaler and DABUS may have lost the battle over the validity of their patent application, they have no doubt won the war to encourage further consideration of the issues surrounding AI, patentability and inventorship – consideration that is only set to become more commonplace and pressing in the coming years.
[1] Thaler v Commissioner of Patents [2021] FCA 879.
[2] Commissioner of Patents v Thaler [2022] FCAFC 62 (Thaler FC).
[3] Reg 3.2(C)(2)(aa) and (4) of the Patents Regulations 1991 (Cth). See Stephen L. Thaler [2021] APO 5 at [34].
[4] Thaler FC at [83].
[5] Thaler FC at [84]-[117].
[6] See Thaler FC at [100]-[101].
[7] See Thaler FC at [107].
[8] Thaler FC at [107]-[110].
[9] Thaler FC at [120].
[10]Thaler FC at [83].
[11] Thaler FC at [120].
[12] Thaler FC at [119].