Introduction
A certain man in 2018 named Dr. Stephen Thaler made a patent application at patent office around the world listing an unusual inventor, DABUS, an Artificial Intelligence system. Dr. Stephen Thaler said the machine had autonomously produced a food container and a flashing light for attracting attention, and Dr. Thaler did not claim the invention himself, then he requested that the machine be named as the inventor. The United States Patent and Trademark Office refused the application; the European Patent Office refused as well, plus the UK Intellectual Property Office’s response was uniformly negative. Courts upheld that the refusal in Thaler v Comptroller-General of Patents [2023] UKSC 49, is that an inventor must be a natural person. These outcomes point how much existing law has fallen behind technology. The Paris Convention and the TRIPS Agreement was drafted when invention was presumed to be done by humans, so they only set the minimum standard for patent protection but had nothing to say when it came to the question about who counts as an inventor. National laws decided to fill in the gap with assumption of human authorship which has led to inconsistency from one jurisdiction to another. This is an issue with serious consequences, because if no one can own AI generated inventions companies may begin to lose the incentive to invest in AI driven research or may even decide to hide AI involvement and give all the credit to a human as the inventor, which undermines the integrity of the patent system. This article is therefore asking if anyone should own AI Generated inventions under International Intellectual property law?
Background/Historical Development:
The patent system has always rested on the idea that invention is a human achievement. The English Statute of Monopolies of 1624 often referred to as the origin of modern patent law is aimed at protecting the ‘true and first inventor of a new manufacture, then further statutes kept this language only on human context and when international rules began in the state this assumption was carried along also. The Paris Convention for the Protection of Industrial Property (1883) was the first major multilateral treaty in this field. This treaty brought about national treatment and priority rights but left each member person free to define patentability. A later revision included article 4ter, which gave the inventor the right to be named in the patent. But, the provision does not define inventor, It only protects the personal link between a person and the invention which is assumed to be human. Copyright law faced the problem first. Some computers started producing music, text & images in the 1960s and 1970s while the Berne convention needed requiring a human author. Some countries responded to this with special rules. In the UK copyright, Designs and Patents Act 1988 Section 9(3) treats the person who makes the arrangements for a computer generated work, the person who gives the prompts and final arrangement of the work as its author. However technology changed. The human contribution has shrunk to just designing the system & selecting data and reviewing results which made things harder and made it harder to say who, if anyone, had authored the output. This pressure got all the way to the international level in the year 2019, when the World Intellectual Property Organization (WIPO) launched its conversation on Intellectual Property Artificial Intelligence and later published a background paper asking if AI-generated inventions should be patentable and who should own them. Just within the space of how special and who should own them. The basics application was filed in 2018. They feared these questions before the next patent worldwide. These universal questions were examined in the next section and confirmed that national law still ties inventorship to natural persons while no international instrument gives a clear answer. And the result is a gap between treaty rules written for human inventors and a technology that now autonomously invents without human directions at every step then leaving ownership uncertain across jurisdictions and for us to close this gap is the central unresolved legal issue for any future international framework.
LEGAL FRAMEWORK:
There is yet to be a single instrument that says who owns an AI-generated invention. Each of the framework is layered. For International treaties, it set standard, the minimum standards, for regional conventions they add procedural rules and for national constitutional and statutes they decide who counts as an inventor and who owns the rights. Constitutional provisions on invention are very rare. So far the best known is Article 1, Sec 8 & Clause 8 of the United States Constitution. It empowers the Congress to secure authors and inventors the exclusive right to their respective writings and discoveries. Some other constitutions have similar provision. Some other constitutions leave patents entirely to ordinary legislation even the US clause is structured around people who create, which has reinforced the common law assumptions that the invention is done by a human. At the international level the TRIPS and Paris Convention provide the baseline. The Article 27(1) of the TRIPS requires patent in all fields of technology without discrimination. Neither does it define inventor or allocates ownership. The Patent Cooperation Treaty required that inventors should be identified in the application but it leaves substantive questions to each designated office, so, ownership is governed by the law of the place where patent protection is sought. Even, national laws out it the same way . The US Patent Act defines the inventor as the individual, who invented the subject matter.
And in Thaler v Vidal the Federal Circuit held that the term Individual means a natural person. In UK Section 7 and 13 of the Patents Act 1977 give the right of a patent primarily to the inventor and it also requires a person to identify them. The Supreme Court held that a machine cannot be an inventor and that owning the machine will not give the owner rights to the output. The Full Federal Court of Australia reached just the same conclusion on appeal. Even the German Federal Court of Justice has demanded that there must be human inventor although it allowed the prompter of the system to be named with reference to its role. South Africa granted Thaler’s application. All together these premises are showing the ownership of a subject matter flows from human authorship and inventorship where a human being makes a large contribution to the idea of the invention. When there is no human meaningful contribution to the idea of the invention. When there’s no meaningful human contribution, there would be no inventor and the invention may fall outside protection. This gap has made policy makers to consider reforms.
Judicial Interpretation and Case Laws
Courts have given the clearest answers on AI inventorship Almost all of the litigation involves one test case the DABUS applications of Dr. Stephen Thaler which named an AI system sole inventor and claimed the ownership of the invention because he was the machines owner. The same fact were argued in some other jurisdiction. The case offers a rare chance to compare how several legal systems read the same statutory language. In the US, the Eastern District of Virginia rejected the application in 2021 and the Federal Circuit confirmed in Thaler v Vidal. The core finding on the Patent Acts definition of an inventor as an “Individual” and relied on the Supreme Court authority that the word literally means natural person. It also noted that the Act refers to an inventor in terms of himself or herself and it left the policy debate to the congress. The Supreme Court declined to hear the case in 2023. The United Kingdom produced the most detailed judicial debate. The High Court and a majority of the Court of Appeal rejected the claim, although Birss LJ dissented, arguing that the Act required only a statement of who the applicant believed the inventor to be. In 2023 as intended Judge of five, the Supreme Court held that the patents Act 1977 treats the inventor as a natural person and that owning a machine does not give a right to its output. It emphasized that it was not dealing with whether AI Generated technical advance deserved patent protection. In Australia it shows how the question was. At first instance, Beach J held that an inventor can be a non-human, reasoning that being that human is a genus noun that can describe a thing as well as a person and that the user of the machine controlled the invention by analogy with ownership of a tree and its fruit. The Full Federal Court overturned that decision in 2022 and held that the Patents Act contemplates a human inventor and the High court refused special leave to appeal. South Africa grants in 2021 is the only contrary outcome anywhere, but the office did not examine the substance of the application. Courts have the following test so being drawn and they decline to stretch it even when they sympathize with the policy arguments. Second, they separated the question of inventorship from that of ownership, since without a legal inventor there is no one from whom title can be derived. Third they repeatedly hand the policy choice back to legislators. And as a result, even the judicial position is consistent but narrow it reveals that machines cannot be inventors.
Critical Analysis:
The correct uniform answer is clear but not quite satisfactory. Taking a critical look would expose the gaps in the common approach, in the arguments for and against change and in the reasoning behind human-only ownership. The strongest of argument for the present position is technical and doctrinal. Patent statutes were written around human inventors and the rights such as being named, assigning and even owning property presuppose legal personality. Granting that status to software would raise hard questions about liability, infringement and succession of title. The courts are right to say that such a major change belongs to the legislature, which can consult widely and weigh effects. Yet the position has its weaknesses. First, it creates a problem of evidence and honesty, because if AI-assisted inventions can only be patented by naming a human then applicants have an incentive to overstate the human contribution or even hide the role of the machine. This undermines the disclosure function of the patent system and makes it harder to know how inventions are made. The German approach of naming a person with a note on the AI’s role is better and more honest and could be widely adopted. Also, the current law leaves the threshold of human contribution unclear. Courts say a person must contribute to conception but they have not explained how much direction, prompting or selection of data is enough. The United States Patent and Trademark Office guidance on AI-assisted inventions applies the joint-inventorship test but some critics say that test fits poorly with a system whose output their users cannot predict the. This uncertainty hurts smaller firms more, since they lack the resources to litigate. Another incentive argument cuts both ways. Supporters of protection say that without patent, firms will under invest in costly AI heavy research, e.g. pharmaceuticals. Opponents respond that AI lowers the cost of invention so sharply that exclusive rights are unnecessary and may be harmful, because they let owners of powerful systems lock up large areas of technology and raise barriers for competitors. Empirical evidence on either side is still thin which weakens confident claims. Plus, the ownership question has been partly addressed because courts have mostly focused on inventorship though co-ownership and several candidates have a plausible claim to own AI outputs, the developer who built the system, the user that prompted it, the owner of the training data and the investor who paid for the computing power. Existing rules say little about how to choose among them and contracts fill the gap, producing different results across platforms. The lack of coordination is a problem because patens are territorial, that is to say, the same invention may be protected in another, which encourages forum shopping and complicates global licencing. The courat overall has been legally correct but practically incomplete. They settled that machines cannot be inventors without settling what counts as sufficient human involvement .
Recommendations
All of the analysis listed above points that the current law,is consistent but incomplete. The following reforms would address it’s main gaps while preserving the human centered foundation of patent law
1) The human contribution threshold should be defined in clear terms: National offices should give out a harmonized guidance on how much human involvement can qualify a person as an inventor. Designing the system for a specific problem and selecting and validating outputs should be able to count so that’s people don’t exaggerate thier roles.
2) Require Use of AI disclosure. Applicants should state the contributions from AI to the invention. This would help with transparency, even the German practices by naming a person with a note on the AI’s role is a workable model.
3. WIPO Coordinate: Members should begin with a nin binding model guideline and consider a later protocol to the patent cooperation Treaty. Since amendment of Treaty is slow, soft law is a more realistic starting point to reduce divergence between jurisdiction.
4. Protect Principle that are human centered : Reforms should keep the requirements that rights belong to accountable legal persons. AI should never hold property or bear liability themselves because by doing so, they would raise enforcement problems that no current legal system is prepared to solve. Out together, these steps would give inventions and investors clearer rules and deeper reforms .
Conclusion
This article sort to find the owner of AI Generated Inventions Under intellectual property Law and wether the framework currently in use can answer the question without reform. International treaties such as Paris Convention and TRIPS say nothing concerning AI inventorship, leaving the matter to national laws. However, courts in the United States, UK, Europe and Australia have applied the law consistently that An inventor must be a natural person and being the owner of a machine does not create rights in what it produces. This clarity is valuable but it is still limited because it does not state how much human input is enough. These gaps create uncertainty, encourages inaccurate inventor designations and produces divergent outcomes even across jurisdictions. The recommendations offered here, offer a measured path for a forward movement, patent law has balanced rewards against access. As AI changes how invention happens there must be a balance struck and it should be revisited with accountability resting on human beings and laid down rules that are clear, transparent and internationally coordinated.
