Artificial Intelligence and Patent Law

As artificial intelligence (AI) continues to advance, the question of whether AI-based inventions are patentable and if AI can be credited as an inventor is gaining prominence in legal and intellectual property circles. These issues are critical in shaping how technology-driven innovations will be protected and commercialized in the future.

Old wine in new wineskins?

The question arises as to whether the key characteristic of AI, the “capability to infer” (cf. the apt specification provided by the European legislator in the AI Act, Regulation (EU) 2024/1689, rec. 12), has an impact on the question of inventorship and the patentability of inventions. The EPO identifies three types of applications of AI in inventive steps (Report from the IP5 expert round table on artificial intelligence, Munich, 31 October 2018): (1.) verification of human-made inventions, (2.) finding the solution after the problem has been identified by a human or (3.) AI-inventions (identification of the problem and solution without human navigation).

The following short review focuses with reference to current case-law on (1.) and (2.) and provides a brief outlook on (3.):

Inventorship?

The complex and highly debated question whether AI itself can be recognized as the inventor was the subject of administrative and court decisions in different jurisdictions in recent years, since inventors attempted to list AI systems as inventors on patent applications. One of the most notable is the "DABUS" case, where the AI system DABUS developed two novel inventions. However, patent offices worldwide, including the U.S. Patent and Trademark Office (USPTO), the European Patent Office (EPO), the German Patent and Trademark Office (GPTO) and the UK Intellectual Property Office (UK IPO), have ruled that only human beings can be recognized as inventors under their current patent law regimes. In Germany and the UK, the GPTO's/UK IPO’s decisions were subsequently confirmed in the last instance by the German Federal Court of Justice in June of this year (decision of 11 June 2024 - X ZB 5/22) and earlier in the UK by the Supreme Court (judgement of 20 December 2023 – Case ID: 2021/0201). These decisions are based on the interpretation that inventorship is tied to a natural person who can have legal standing, responsibility, and intent. And only natural persons can make an invention through a creative act. AI, as it currently stands, lacks these legal and cognitive attributes. While AI systems may generate inventive outputs, the respective patent law regimes maintain that a human must have contributed to the conception of the invention or its application.

Patentability?

AI inventions often fall under the categories of software, algorithms, or machine learning models. In these cases, the patent offices typically assess whether the AI system offers a tangible technological improvement over existing processes or tools. For example, AI-driven systems in healthcare diagnostics, autonomous vehicles, or manufacturing can be patented if they introduce a new and inventive process and are tied to a technical field. Therefore, AI-based inventions can be patentable, provided they meet the standard criteria for patent eligibility. An example of maintaining the current standards is the noted decision of the UK Court of Appeal in the Emotional Perception AI's case, where the court decided that the patent application for a system using an artificial neural network (ANN) to recommend and organize data files should be treated like any other computer program. ANN technology fell under the exclusion from patentability as a computer program, but such innovations could still be patentable if they make a technical contribution.

Within the conflicting areas between the function of AI and human actions, the inventive step may be the key criterion for determining the entitlement to the invention. Both, the EPO and the German case law follow an objective approach for determining the inventive step. This objective approach assumes that the AI is merely a tool of the human inventor – either by instructing the AI to find the solution or by recognizing the solution provided by the AI as such. As a result, inventive step may also result from the mere “realization” of AI-generated teachings and solutions. However, the question whether a natural person will also be entitled to file an application could arise in particular in the event that AI operates completely autonomously in the search for (a problem and) a solution and the results are no longer attributable to humans. De lege lata, a court might have to come to the conclusion that the invention is in the public domain, since there is no inventor, unless it suffices under case law that the (entitled) inventor does not have to make any inventive step at all when finding the solution.

Outlook

As AI continues to evolve, there is growing pressure to rethink how intellectual property law and especially patent law addresses AI-driven innovations. The current framework, which restricts inventorship to humans, could eventually require amendments as AI becomes more autonomous in creating inventions with decreasing human contributions. While AI cannot yet be recognized as an inventor, the discussion is ongoing, with some advocates pushing for legal reforms that would allow AI to hold inventorship status or co-inventorship with humans. As AI’s capabilities expand, legal scholars, patent offices and courts will continue to grapple with the implications for patents, potentially reshaping the landscape of innovation protection.