AI as inventor: Legal challenges and implications for patent law

In the rapidly evolving landscape of our world, AI has progressively taken on a central role, with its capacity

to facilitate inventions becoming undisputed.

This fascinating development necessitates some critical legal considerations, especially concerning the patent system. This article explores some of the most contentious issues, including recognizing AI as an inventor, owning moral and economic rights resulting from AI-related inventions, and the potential challenges and advantages of extending patent protection to inventions autonomously generated by AI.

Case law on the qualification of AI as an inventor

Regarding the qualification of AI as an inventor, the European Patent Office (EPO) recently ruled on the issue, rejecting two patent applications designating an AI system called DABUS (Device for Autonomous Bootstrapping of Unified Sentience) as an inventor. These applications were filed by DABUS creator, Steven Thaler, as part of the Artificial Inventor Project initiative, which aims to claim intellectual economic rights for inventions generated by AI in the absence of human creative input.

In particular, EPO considered that the designation of DABUS as an inventor was contrary to the European Patent Convention, which in its opinion would presuppose that the inventor is a natural person. The Office also considered that the applicant could not even be considered as the successor in title of the machine, since the latter has no legal personality and cannot transfer any rights.

Similar conclusions have been reached by other Offices, notably the US Patent Office (USPTO) and the UK Patent Office (UKIPO), while DABUS patent applications have succeeded in South Africa and Australia, where the possibility of recognizing AI as an inventor has become a reality.

The ownership of the moral and economic rights on the invention generated by AI

It’s clear that to date there are no unanimous decisions on the matter, and there are many aspects particularlydebated, starting with the ownership of moral andeconomic rights on the invention generated by the AI.

As for moral rights, considering that they traditionally belong to the inventor, they should be, by default, owned by the machine. However, it is not self-evident that an AI system can claim moral rights in the absence of legal personality. At the same time, it would be necessary to determine who is entitled to the economic rights arising from the patent; to this end, the candidates could include the owner of the machine, its user, the programmer or, again, the person who developed the algorithm.

The effects of extending patent protection to AI-generated inventions

The idea of extending patent protection to AI-generated inventions carries numerous implications.

For instance, satisfying the legal requirements for patentability, particularly novelty and inventive step, may become more challenging. Novelty, as defined by the law, refers to an invention that does not already form part of the state of the art when the application is filed.

With AI having access to an ever-increasing amount of information and thereby broadening the scope of the state of the art, assessing an invention’s noveltycould be more difficult. At the same time, however, itcould be argued that mere access to more informationdoes not automatically imply a greater state of the art,since it is also necessary for the recipient to be able to understand such disclosed knowledge.

With regard to the inventive step requirement, according to which an invention does not have to be obvious from the state of the art to a person skilled in the art, in view of the greater breadth of information available at the time of filing the application, it may be necessary to rethink the person skilled in the art parameter in the case of inventions generated by the AI, commensurate with the AI’s skill.

Equally challenging would be the assessment of the sufficiency of disclosure. In fact, the patent system relies on a sort of agreement between the inventor and society: the possibility to exclude competitors must be balanced by the benefit that people derive from the clear and complete disclosure of the features of the invention, to enrich the common scientific heritage and, once the patent monopoly has expired, allow the exploitation of the invention. However, AI creations may represent the epilogue of an inventive process that is not necessarily intelligible to human beings, with the consequence that their sufficiently clear and complete disclosure may not prove to be a particularly easy task.

The risks of not recognizing the patentability of AI creations

On the other hand, not recognizing the patentability of AI-generated creations could have drawbacks. For example, there could be a risk that the invention is nonetheless generated by AI, but that this is not declared, and people not involved in the inventive process take credit for it. In addition, denying patent protection to inventions created by AI could lead to greater recourse to the protection provided by trade secret rules and, therefore, less sharing of knowledge with the broader community.

In conclusion, there is certainly a question as to whether patent law is currently adequately equipped to answer the many questions that this issue raises. The hope is that the important implications that technological development brings with it will be taken into consideration, also in the context of a political debate, so they can find an adequate regulatory response.