The fundamental recognition of IPR has been adopted in the International Covenant on Economic, Social, and Cultural Rights (ICESCR), which recognizes two rights- the right of the general public to enjoy the benefit of scientific progress; and the protection of moral and material interests of the person producing such progress. Article 4 of the Paris Convention specifies the right of the inventor to be named in the patent and also the right of priority. In 2021, during the ongoing debate on whether AI can be recognized as an “inventor”, the Australian Court, in Thaler v. Commissioner of Patents, has held affirmative, contrary to the positions held by the United States and European Patent Office. Therefore, this article shall briefly analyze the laws and precedents in the United States, the EU, and India and their ability to include AI within their regimes.
United States Of America (USA)
In the United States, the patent law has adopted the utilitarian approach by basing the law’s ideology that utility can be achieved by encouraging or incentivizing “human” inventors“.
The U.S. patent law does not mandate a particular level of control or input by humans, but the law frames questions of inventorship and patentability in terms of human creation and requires conception. Such conception has been applied to include the formulation of ideas in the mind of the inventor, reflected in the case of Sewall v. Walters. This is reflected in the wording “whoever” and “person“, which operates in conformity to the approach of the Copyright Office of not granting protection to AI-generated works. Additionally, Section 115 requires that the correct inventor(s) be named in a patent application.
The landmark case Townsend v. Smith emphasises the importance of ‘conception’ for an outcome to be construed as valid. This case indicates that the invention must have been conceived (a definite and complete idea of the operative invention) in the mind of the inventor before it can be materialised and patented. Such a requirement poses difficulty in granting patents to AI; or requires the legal systems to make an exception for AI.
The International Tin Council case allows one to understand that any electronic personality would award a defence to human beings for violation of any rights of any human or corporate agency.
In the case of Alice Corporation v. CLS Bank International, the decision of the apex court of the United States created a setback for the patent laws for many years. Holding that an abstract idea on a computer that aids in settlement risk is not sufficient to obtain a patent, thus invalidating any efforts taken to recognize AI as a patent owner. The only solace that can be sought in the entire judgement was the opinion of Judge Newman, who called for more clarity on the enforcement of low-quality software patents.
In the case mentioned above, Free Software Foundation, acting as amicus curiae, delivered an interesting opinion that the test to determine whether an invention by artificial intelligence would fall within the ambit of the machine-or-transformation test.
In Acohs Pty Ltd v Ucorp Pty Ltd, an Australian Court declared that a work generated with the intervention of a computer could not be protected by copyright because it was not produced by a human. The same can be extended to patent law, which requires the mental element to be present.
In Schillinger v. United States, decided by the United States Supreme Court, holding that a suit for infringement of a patent cannot be considered against the United States because an infringement of a patent is a tort, and the United States has not surrendered sovereign immunity for intentional torts. This judgment should be considered for the purpose of infringement. How would the legal systems address the liability of AI in matters of infringement of patents of others, particularly other AI systems?
An analogue can be drawn from the discussion of ‘person’ in copyright law. It is pertinent that a human has to be an author for the purposes of the creation of the work, as noted in Naruto v. Slater. The courts have also refused to issue writs to chimpanzees on the grounds that they are incapable of being held legally accountable for their actions, the incapability to bear any legal responsibilities and societal duties. This legal reasoning must be extended to AI as well.
USA has addressed the concern inherent in machine-generated works, noting that a determination of the line between human and machine participation would be a “crucial question”. As noted, the U.S. framework suggests that finding the line between whether an AI was merely a tool or whether an AI independently conceived and executed work is difficult.
European Union (EU)
The European Union has also adopted a similar approach. European Patent Convention (EPC), under Article 58, provides that an application can be filed by a natural or a legal person. However, the article allows for applicants (those equated to legal status by the law governing), whereby AI can be allowed as long as the law grants the status. However, no such grant has been afforded so far, rendering the alternate inapplicable. Article 60(1) EPC is concerned with safeguarding the proprietary interest of the inventor, and any introduction of AI will require legal systems to discuss the interests of the programmers of the AI as well.
EU has commissioned reports which recognise the Asimov Rules and require legal systems to initiate structured public dialogues to discuss the consequences of the technology. The report also raises the possibility of applying existing legal principles to AI with amendments. The AI Experts’ Open Letter to the EU Commission on AI Robotics rejects the need to establish legal personhood for AI systems and attempts to award legal personality steming from a superficial understanding of the technological and legal landscape in which AI systems operate.
Direct recognition of intellectual property law is found in the 7th schedule of the Indian Constitution. Indian Patents Act, 1970 forms the primary consideration as Section 2(1)(j) defines “invention”, which is required to be re-evaluated (in light of the judicial decisions) which require a ‘mental act’ element; whether the law can eliminate this requirement for AI-generated inventions (AI as first and true inventor). Section 6 is an important provision for evaluation as it addresses requirements to apply for patents. The provision employs the term “person”, which extends to entities (Rule 2) as well. However, the law will also have to respond with additional incentives for AI, in terms of legal responsibility and enforceability of rights.
Section 3(k) provides that the computer program or algorithm cannot be patented. This section raises red flags for the final output as AI programming language is mainly algorithm-based (whether the output of a non-patentable algorithm is eligible for a patent).
For an invention to be patentable, it is necessary that the invention must be a new product or process and useful. In Bishwanath Radhe Shyam v. Hindustan Metal Industries, it was held that the step must not be a mere workshop improvement but new and useful. The primary question involves developing criteria to qualify ‘new’ inventions for AI.
To seek injunctions to prevent infringement, the patent holder must show he would suffer irreparable injury if the injunction was not granted. This judgement would increase the burden of proof on AI to show significant injury, and yet the courts will have to evaluate if such consideration is sufficient to grant an injunction.
In Franz Xaver Huemer Vs. New Yash Engineers, the court noted the negative impact on markets and the economy if patents were granted and not utilised for commercial purposes. The exclusive right granted to the inventor over his invention includes his right to assign his invention or transfer any interest in his invention in favour of any other person in consideration of monetary gain. How can AI negotiate the terms of usage regarding the patent? Moreover, will this require a need to reconsider the existing contract law.
Additionally, patent law requires full disclosure of the steps involved; it is pertinent that the market has actual knowledge of the invention, and not just the benefits conferred. The programming language of AI can pose an obstacle to such disclosure.
As more machines are being invented, and if they are possess the same capabilities as DABUS, the current patent laws become unfit. The existing laws allow for juristic persons to be inventors considering that legal entities are capable of enforcing rights and liabilities and can be held socially accountable. Additionally, patent laws are territory specific, and the lack of uniformity in the treatment of artificial intelligence distorts the legal systems of other countries that have not recognised AI as persons, as no protection will be afforded in such nations.
Laws across jurisdictions apply the obviousness test based on human capabilities assessment. Including AI within the meaning of ‘person’ will require legal systems to relax or waive the test, potentially increasing the number of frivolous patent applications.
Some jurisdictions require conception to be made in relation to the actual language of the claims; others define “invention” in a broader sense as to include additional material in the patent application.
Recognition of AI will require a re-evaluation of the right to attribution. This right largely serves two objectives- to inform the public and then allow the public to appreciate the invention. Thus, in view of the ongoing discussions in AI-related laws, the underlying objective is incentivising – how can the legal system incentivise AI to disclose innovations?