Back in December 2020 I reported on the progress of an Australian patent application naming an ‘artificial intelligence’ going by the name DABUS (‘Device for the Autonomous Bootstrapping of Unified Sentience’) as inventor. The application – no. 2019363177 – was derived from an international application filed under the Patent Cooperation Treaty (PCT), and names Dr Stephen Thaler, the developer of DABUS, as applicant. At that time, the Australian Patent Office had raised an objection to the naming of a machine as inventor, and sought further submissions from the applicant on this issue. I anticipated that the case would be referred to a hearing officer to determine the matter by way of a formal decision, and predicted that Dr Thaler’s rather unconventional assertions that an AI can be an inventor, and that he is entitled to receive any resulting patents by virtue of ownership of the AI, were unlikely to be accepted by the Patent Office.
That has all now come to pass, with a decision of a delegate of the Commission of Patents issued on 9 February 2021: Stephen L. Thaler [2021] APO 5. In accordance with the plan language of the Australian Patents Act, and established legal principles, the delegate has determined that a patent can only be granted to a person, that a machine is not a person, and that there is no mechanism whereby the right to apply for a patent on an invention devised by a machine can be legally transferred to any person.
In short, a patent application naming a non-human inventor cannot be validly filed, regardless of who is named as the applicant. A corollary of this (although not addressed in the decision) is that a patent on an invention devised by a machine can never be validly granted, even if the true nature of the inventor is concealed from the Patent Office, because there is simply no mechanism for any named applicant to derive a legal right to apply. In practice, applicants are required to provide the Patent Office with a declaration of entitlement, and in the case of a machine invention this declaration would necessarily be invalid and/or false.
Can Machines Really ‘Invent’?
The Australian Patent Office decision, in common with similar decisions issued at the European Patent Office, in the UK, and in the US, does not address the question of whether a machine is actually capable of invention. The applicant has asserted that the inventor is DABUS, explicitly stating in the application that ‘the invention was autonomously generated by an artificial intelligence’. The usual practice of the Patent Office, in the absence of evidence to the contrary, is to accept an applicant’s identification of the inventor at face value. The only question, at this point, is whether a non-human inventor satisfies the requirements of the Patents Act.
For what it is worth, my own view remains that machines are not currently (and may never be) capable of invention (for more on this, see The Impact of Machine Learning on Patent Law, Part 1: Can a Computer ‘Invent’? and Free Online Seminar – Machine Inventors, Fact or Science Fiction?) While this raises a substantive question of who (if anyone) is actually the inventor in the DABUS case, this falls beyond the scope of the present decision.
Patents Cannot be Granted to Machines
Subsection 15(1) of the Patents Act 1990 provides that:
Subject to this Act, a patent for an invention may only be granted to a person who:
(a) is the inventor; or
(b) would, on grant of a patent for the invention, be entitled to have the patent assigned to the person; or
(c) derives title to the invention from the inventor or a person mentioned in paragraph (b); or
(d) is the legal representative of a deceased person mentioned in paragraph (a), (b) or (c).
Key to the operation of this provision is that the recipient of a patent must be a person. Subsection 2C(1) of the Acts Interpretation Act 1901, which applies to the Patents Act, provides that ‘[i]n any Act, expressions used to denote persons generally (such as “person”, “party”, “someone”, “anyone”, “no-one”, “one”, “another” and “whoever”), include a body politic or corporate as well as an individual.’ But while the law grants legal personality to a ‘body politic or corporate’, as well as to actual (i.e. ‘natural’) persons, there is nothing in the Patents Act, the Acts Interpretation Act, or the common law, that would allow a machine to be treated as ‘a person’ for the purposes of being granted a patent. Furthermore, the categories of permissible patentees established by subsection 15(1) are exhaustive. Unless one (or more) of the four condition is met by an applicant, they are not entitled to be granted a patent.
Paragraph 15(1)(a) clearly cannot apply to a machine. As explained at paragraph [20] of the Patent Office decision:
… the entitlement of the patentee flows from the inventor, and absent devolution the inventor will become the patentee. That appears to imply that the inventor must also be a person. An alternative interpretation of section 15(1) is that an inventor who is not a person cannot become a patentee, and the patentee must instead be a person who satisfies either paragraph (b), (c) or (d).
The first interpretation, i.e. that the scheme of the Act requires that only a natural person can be an inventor, is essentially the conclusion reached by the England and Wales High Court in relation to the corresponding provisions of the UK Patents Act (Thaler v The Comptroller-General of Patents, Designs And Trade Marks [2020] EWHC 2412 (Pat), on which I reported back in September 2020). This is also, as it happens, my own preferred interpretation. If correct, that would be the end of the matter, because it means that an invention devised by a machine – or, indeed, in any way that does not involve some intellectual exertion by an actual person – simply cannot be the subject of a valid patent.
The delegate went on, however, to consider the alternative proposition that a person might satisfy one of paragraphs 15(1)(b) or (c) (or, briefly, (d), which obviously does not apply here), even if paragraph 15(1)(a) cannot apply. This would, of course, require such a person to acquire title to the invention, and the patent/application, in one of the ways contemplated by the Act.
Machines Cannot Assign Property
The delegate dealt swiftly with paragraph 15(1)(b), which is based upon a prospective assignment of a granted patent to the nominated applicant, noting (at [26]) that:
It is an uncontroversial observation that the law does not presently recognise the capacity of an artificial intelligence machine to assign property. Consequently section 15(1)(b) is not consistent with treating an artificial intelligence machine as an inventor.
Personally, I would go further than this. Since a machine is incapable of owning or assigning property, paragraph 15(1)(b) surely cannot apply even if the machine could be treated as an inventor – there is no need to appeal to ‘consistency’. Either way, however, the provision cannot operate to entitle a person (natural or otherwise) to be granted a patent on an invention allegedly devised by a non-person.
Can an Applicant ‘Derive Title’ from a Machine?
The applicant did not, in any event, seek to rely upon paragraph 15(1)(b). His argument (at, e.g., [29]) was that paragraph 15(1)(c) applies because he ‘derives title to the invention from the inventor’ by virtue of his ownership of DABUS:
The common law confers ownership of anything produced by DABUS to its owner, the applicant. There is a general rule that the owner of a thing is the owner of the fruits of that thing, much like the owner of a fruit tree is entitled to the fruit produced by that tree. The principle of accession or first possession can apply.
Notably, the applicant did not attempt to argue that DABUS has legal personality, nor that it is capable of transferring or assigning intellectual property, nor that it is even capable of owning property in the first place. Put another way, DABUS is a ‘pure’ autonomous generator of intellectual property, possession of which passes automatically to DABUS’ owner, Dr Stephen Thaler.
The delegate was not persuaded by this argument (at [30]):
The principle of first possession states that a person who takes possession of something that is ownerless is entitled to ownership simply by possession. The doctrine of accession seems more relevant to the present case, but the following analysis is the same in both cases. For present purposes, the critical question posed by section 15(1)(c) is whether ownership that is created by accession (or possession) is derived “from the inventor”. It seems that ownership automatically vests in the owner of the primary property by virtue of their ownership of that property (or possession of the new property) and is not regarded as transferred from another person. While the principle of accession (or possession) might well deal with the ownership of an invention created by an artificial intelligence machine, it does not do so by conceptually moving title "from" the artificial intelligence machine to the owner of the machine. It follows that the principle of accession (or possession) does not provide a means to identify a person who can be granted a patent that is consistent with section 15(1)(c).
There is, I think, considerable force in this reasoning, which can perhaps be emphasised by circling back to the fruit tree analogy. While I am the owner of the limes that grow on the tree in my back garden, most people would think it rather odd if I were to say that I ‘derive title to the limes from my lime tree’. Clearly this is not accurate. I simply own the limes, just as I own the tree. There never was any ‘title’ to the limes that was held by the tree, let alone a transfer of title to me.
‘Communication’ of Inventions
The delegate also considered whether an invention could be ‘communicated’ from a machine to an applicant, according to the specific meaning of that term under earlier patent law (i.e. the Patents Act 1952), and on the basis that this mechanism is now subsumed into paragraph 15(1)(c). This possibility has never been argued by the applicant. It was first raised by the Patent Office, in correspondence to the applicant’s attorneys dated 25 November 2020, and I have discussed it previously at some length. My own view – notwithstanding obiter dicta of a Full Bench of the Federal Court of Australia in JMVB Enterprises Pty Ltd v Camoflag Pty Ltd [2006] FCAFC 141 – is that the concept of ‘communication’ has little, if any, application under the current Act. Regardless, I observed in my earlier article that a ‘communicatee’ seems to have been a kind of trustee, obliged to act in the best interests of, and having a fiduciary duty to, the ‘true’ inventor or assignee, who retains a beneficial interest in the patent application.
The delegate appears to have reached a similar conclusion, observing (at [28]) that:
… it is inconsistent with my understanding of the law that an artificial intelligence machine can have a beneficial interest in any property. It follows that I cannot see how the owner of a machine can be regarded as the communicatee of an artificial intelligence machine.
Conclusion – The Right Outcome
In summary, the Patent Office has determined that the Patents Act 1990 does not permit a machine to be named as an inventor and, indeed, inherently requires there to be an inventor who is a natural person. Specifically, the scheme of subsection 15(1) requires the inventor to be a person capable of legally owning and transferring property. A machine, such as DABUS, lacks legal personality, and can do neither of these things.
There is an interesting – though distasteful – historical antecedent for this situation, which I wrote about way back in February 2013. In pre-Civil War America there were slave-owners, such as future Confederate President Jefferson Davis and his brother Joseph, who considered that they were entitled to the fruits of the ingenuity of their human chattels, and should be able to obtain patents in their own names in respect of their slaves’ inventions. However, the US Commissioner of Patents, Joseph Holt, disagreed. In 1857 he ruled that since slaves had no legal standing, they could not swear the oath of inventorship required under the US patent law. This presumably remains a problem in the US, where the execution of an oath or declaration by the named inventor(s) remains a requirement, although the USPTO has thus far rejected Thaler’s attempt to file an application for the DABUS inventions on the more fundamental ground that an inventor must be a natural person. Equally, a lack of legal standing for slaves to own, sell, and otherwise transfer property would have created a further barrier to slave owners’ claims on the inventions of their slaves.
I obviously do not intend to equate DABUS’ situation with human slavery. DABUS is a machine. It has no human rights, and it does not experience oppression or suffering (unless you choose to believe some of Dr Thaler’s more outlandish claims about ‘brain trauma and death’ or ‘mental illness’ in artificial neural networks). My point is simply that there is nothing new under the sun: people have previously tried – and failed – to claim ownership of inventions on the basis of ‘ownership’ of the inventor. The law has not relevantly changed in the meantime.
In 1850’s America, the proposed solution – thankfully never enacted – was to amend the patent laws to expressly recognise the rights of slave owners to apply for and obtain patents in respect of inventions devised by their slaves. That would have been wrong back then, and an analogous change – which is what DABUS’ backers are really looking to achieve – would be wrong now.
All natural persons are free to invent, and should have equal rights to seek protection for their inventions under whatever conditions the law prescribes. While the pathway from inventor to applicant has historically been more expansive than it is today, embracing mechanisms such as importation and ‘communication’, the right to apply for, and to obtain, a patent has never ultimately derived from anywhere other than the creativity and ingenuity of a human mind. A persuasive case has not be made out for any change in this position. Indeed, I have argued elsewhere that allowing patents for inventions churned out by tireless machines with virtually unlimited capacity, without the further exercise of any human ingenuity, judgment, or intellectual effort, may simply incentivise large corporations to build ‘patent thicket generators’ that could only serve to stifle, rather than encourage, innovation overall.
The Australian Patent Office, in line with the European Patent Office, the UK Intellectual Property Office, the England and Wales High Court, and the USPTO, has reached the right decision. In those other jurisdictions, Thaler has appealed the patent office decisions, and now has the right to appeal to the Federal Court of Australia. If that happens, then I confidently predict that the outcome here will be just the same.
0 comments:
Post a Comment