USPTO and EPO reject the notion of AI as an inventor, while the established opinion in the copyright world is that an AI machine cannot generate copyrighted works. In the article at hand, we will present the recent case law regarding IP rights and AI machines and we will analyze the core argument that connects the refusal of inventorship to AI machines.
Artificial intelligence was publicly noticed by people through movies such as “2001: A Space Odyssey” and “I, Robot”; fascinating – and a little scary, AI was an entertaining feature that was considered science fiction. Nowadays, copyright and patent law lectures include discussions on whether AI could be an author of a copyrighted work or whether patent offices should accept AI as an inventor. These discussions were mostly based the relevant statutes– not technologically neutral though, since there were no related decisions by Courts or patent offices.
During the last six months, both the European Patent Office and the United States Patent and Trademark Office have rejected patent applications where the inventor is an AI system. More specifically, at the end of April, the USPTO denied vacating a Notice of Missing Parts for U.S. Patent Application No. 16/524,350, because the listed inventor was not a person but DABUS, an artificial intelligence system (“failure to identify each inventor by his or her legal name”).
The applicant of said patent application argued that DABUS was the rightful inventor, since it managed to“independently create this invention” and there is no specific mention in the US statute for an inventor to be a natural person. Conversely, the USPTO held that, even though the explicit term “natural person” is not in the text of the US Patent Act, the inventor is addressed with pronouns that are used for natural persons such as “whoever” (35 U.S.C. § 101) or “himself”, “herself” and “individual” (35 U.S.C. § 115).USPTO cited Federal Circuit case-law (Univ. of Utah v. Max-Planck-GesellschaflzurForderung der Wissenschaflene.V.) that refers to the inventor or inventors as “individual or groups of individuals”. Finally, USPTO stated that the interpretation of the term “inventor” in such a broad fashion to include an AI system “would contradict the plain reading of the patent statutes that refer to persons and individuals”.
On the other side of the Atlantic, EPO denied the granting of EP 18 275 163 and EP 18 275 174 – two European patent applications for “a food container” and “devices and methods for attracting enhanced attention” – that indicated DABUS as a single inventor. EPO stated that the interpretation of the European Patent Convention leads us to consider that the inventor must solely be a person, a human being and not an AI system, a machine. EPO further clarified that this construction of the term “inventor” is internationally acceptable; also, there are no legal statutes that establish a machine as an inventor.
Nevertheless, EPO took one step further, highlighting that natural persons have legal personality that enables them to have and exercise their rights; lack of legal personality equals inability to have rights. Since machines – and in our case AI systems – are no persons, they have not legal personality and cannot be inventors. Moreover, giving the AI system a name does not meet the requirements of Rule 19(1) EPC. EPO argued that the name does not only serve an identifying purpose for the person, but also “enables them to exercise their rights”; machines that have no rights with or without a name. The same decision and rationale was provided by the UK Intellectual Property Office during the examination of the corresponding applications; DABUS, being a machine, cannot hold property, thus it cannot be an inventor.
But if we set aside the statutes and majority opinions for a second, it is apparent that we are facing more than just patent applications listing an inadmissible inventor; we are facing the future and there are certain policy considerations that we should have in mind. The USPTO decided to neglect these considerations that were brought before it by the applicant and stated, “they do not overcome the plain language of the patent laws as passed by the Congress and as interpreted by the courts”.
On the other hand, Hearing Officer Jones of UKIPO addressed these considerations in the conclusion of the decision; more specifically, he claimed that the patent protection functions as a reward for the inventor in order to incentivize her to innovate further and this function is not applicable to AI, since “an AI machine is unlikely to be motivated to innovate by the prospect of obtaining patent protection”. Even though AI systems do not have the incentive to invent as human beings do, this doesn’t mean that they will stop inventing. Patent law cannot hide forever behind legal provisions and international consensus; Officer Jones continued in the UKIPO decision to say, “there is a legitimate question as to how or whether the patent system should handle [AI] inventions”.
Indeed it is understandable that, as USPTO made clear, policy considerations cannot surpass the law and that the existing system cannot encompass machines in the term “inventor”, but that does not mean necessarily that the existing term or the existing system should remain the same forever. Technology is moving rapidly and the law has to follow. If it doesn’t, brace yourselves for a lot more patent applications where the listed inventor is improperly named a natural person in order to avoid the patent office rejections. Turning a blind eye to this practice that the applicants have adopted is not making the patent offices or the public any favors; to the contrary, no respect is paid to the right of the public to be informed about the identity of the real inventor and to the real inventor whether they are human or machine.