When Algorithms Become Inventors
The relationship between innovation and intellectual property has always evolved alongside technological change. The industrial revolution reshaped patent law for mechanical inventions, the digital era forced courts to tackle software patents, and today, artificial intelligence is raising perhaps the most fundamental question yet: can a machine be an inventor?
Modern research environments increasingly rely on machine learning systems capable of generating new molecules, optimising engineering designs, and discovering materials without direct human intervention. Generative AI models are already transforming pharmaceutical discovery, climate technology development, and industrial design. In many laboratories, algorithms are not merely assisting scientists but actively generating novel solutions that human researchers later refine.
This shift has triggered an intense legal debate across jurisdictions. Patent law has traditionally assumed that inventors are human beings who exercise intellectual creativity. Yet when AI systems autonomously generate patentable inventions, the law faces a conceptual challenge. If the inventive step originates from an algorithm rather than a person, who should legally be recognised as the inventor? Complicating matters further is the way many AI systems learn and generate ideas: by training on vast volumes of existing online content, research papers, patents, and creative works. This raises an additional question: when an AI-derived invention draws from this ocean of pre-existing knowledge, could it inadvertently cross into the territory of copyright infringement or unauthorised use of protected intellectual property?
A compelling counterview is that AI systems do not truly ‘invent’ in the legal sense, but rather synthesise and recombine existing human knowledge embedded in their training data. If the underlying inputs, objectives, and interpretive frameworks are human-defined, the inventive contribution may ultimately remain traceable to human ingenuity, challenging the premise of autonomous AI inventorship
The debate is not merely philosophical. It has practical implications for innovation ecosystems, investment decisions, and global competitiveness. For countries seeking to position themselves as leaders in deep-tech innovation, clarity around AI-generated inventions is rapidly becoming a policy priority.
The Global Legal Debate: Lessons from the DABUS Cases
The most prominent legal battle over AI inventorship arose from global litigation over the AI system known as DABUS (Device for the Autonomous Bootstrapping of Unified Sentience). It aims to establish that an AI system can be recognized as an inventor on a patent application. Developed by computer scientist Stephen Thaler, DABUS allegedly generated two inventions independently: a fractal container design and a flashing emergency beacon. Thaler filed patent applications in multiple jurisdictions listing DABUS itself as the inventor. As the creator and owner of DABUS, Dr Thaler claimed he was entitled to the patents, as the ownership rights flowed from the AI to him.
The resulting legal proceedings triggered one of the most widely discussed intellectual property debates of the decade. Courts and patent offices were forced to interpret statutory language written long before autonomous AI systems existed. Across most jurisdictions, including the US, UK, and Europe, the legal conclusion was consistent: under current laws, only a human being can qualify as an inventor. However, the reasoning behind these decisions reveals deeper tensions between technological reality and legal doctrine.
United States: The Natural Person Requirement
In the United States, the applications were rejected by the United States Patent and Trademark Office. The decision was later affirmed by the United States Court of Appeals for the Federal Circuit, which held that the term “inventor” under the US Patent Act refers to a natural person.
The court’s reasoning relied heavily on statutory language referring to inventors using pronouns such as “himself” or “herself.” According to the court, this linguistic structure reflected legislative intent that inventorship must be limited to human beings. The Supreme Court of the United States declined to review the decision, effectively closing the door on AI inventorship under current American law1.
Europe and the United Kingdom: Legal Formalism vs Technological Change
A similar outcome emerged in Europe. The European Patent Office rejected the DABUS applications, emphasising that the European Patent Convention requires inventors to possess legal personality. Because AI systems lack legal rights and obligations, they cannot satisfy this requirement.
The same reasoning prevailed in the United Kingdom. The UK Intellectual Property Office refused the application, a decision later upheld by the Supreme Court of the United Kingdom in 2023. The Court concluded that the UK Patents Act assumes an inventor must be a natural person and that ownership rights flow from that human inventorship.
These decisions illustrate a broader judicial tendency: rather than reinterpret statutory language to accommodate AI inventors, courts have chosen to defer to legislatures.
South Africa’s Unexpected Divergence
One jurisdiction diverged from this global consensus. In 2021, South Africa granted a patent listing DABUS as the inventor through the Companies and Intellectual Property Commission2. However, the decision did not involve a substantive examination of inventorship requirements because South Africa operates a largely formal registration system.
While symbolically significant, the decision does not necessarily indicate a broader acceptance of AI inventorship. Instead, it highlights how procedural differences between patent systems can produce divergent outcomes.
India’s Current Position under the Patents Act
India has already engaged with questions surrounding AI inventorship at the Patent Office level, although the issue has not yet been conclusively tested before Indian courts. In a decision issued by the Controller, the Indian Patent Office aligned itself with the prevailing global approach by reinforcing the requirement of human inventorship under the existing patent framework.
The governing legislation, the Patents Act, 1970, defines eligibility for filing patent applications under Section 6. The provision allows applications by the “true and first inventor,” an assignee, or a legal representative of the inventor.
While the statute does not expressly define ‘inventor’ as a human being, the broader legislative scheme and prevailing patent practice interpret the term as referring to a natural person who contributes to the inventive concept. Consequently, under the current framework, an AI system cannot independently be recognised as an inventor in India.
At the same time, India’s innovation ecosystem is rapidly expanding in areas where AI-generated inventions are becoming common. Deep-tech start-ups, pharmaceutical research institutions, and climate technology companies increasingly rely on AI-driven discovery platforms. This creates an interesting paradox. India’s legal framework implicitly assumes human inventors, yet its innovation economy increasingly depends on AI-assisted research.
India’s legal position on AI-generated creations remains cautious and evolving. While existing intellectual property laws were designed for human creators, recent developments, including the brief recognition and subsequent reconsideration of AI co-authorship in the RAGHAV case3, illustrate the regulatory uncertainty surrounding AI-generated works. Policymakers increasingly recognise the need to update the legal framework to balance innovation with the protection of creators’ rights as AI-driven innovation accelerates.
Practical Challenges for Patent Offices in the Age of AI
Even if legislators were willing to reconsider inventorship rules, patent offices would face significant operational challenges in examining AI-generated inventions.
One issue concerns disclosure requirements. Patent law requires applicants to fully disclose how an invention works so that others skilled in the art can reproduce it. However, many advanced AI systems function as “black boxes,” producing outputs without transparent reasoning processes. When an invention emerges from complex neural networks trained on vast datasets, explaining the inventive step becomes far more difficult.
Another challenge relates to accountability. Patent systems rely on inventor declarations and legal responsibility for the accuracy of disclosures. AI systems cannot assume legal liability or provide sworn declarations. This raises questions about who should bear responsibility for errors or misconduct in AI-generated patent applications.
A related issue is the increasing complexity of invention disclosures themselves. In-house patent attorneys already struggle with evaluating highly technical research outputs generated by interdisciplinary teams. AI-generated innovations further complicate this process by blurring the line between human contribution and algorithmic discovery.
Policy Pathways for India’s Patent System
For policymakers, the question is not simply whether AI should be recognised as an inventor. Instead, the more practical challenge is designing a system that encourages innovation while preserving legal certainty.
One possible approach would be to maintain the current human-inventor requirement while allowing AI-assisted inventions to be attributed to the human researchers who designed or directed the system. This approach mirrors current practice in most jurisdictions and avoids the need for major statutory reform.
Another option would involve creating a new category of inventorship recognising AI-generated contributions while assigning ownership to human operators or organisations. Such a model could preserve accountability while acknowledging the technological realities of modern research.
A more radical possibility would be to develop a completely new intellectual property framework for AI-generated inventions. However, this approach would raise complex questions about ownership, liability, and enforceability.
For India, any policy decision must consider broader national priorities. The country is investing heavily in artificial intelligence, climate technologies, and pharmaceutical research. Clear IP rules could play a significant role in attracting global R&D investment and supporting domestic innovation ecosystems.
Strategic Implications for R&D Leaders and Innovators
While legislators debate future reforms, companies cannot afford to wait for legal clarity. Organisations using AI-driven research tools must already have strategies in place to protect their intellectual property.
The most practical approach today is to ensure that human researchers remain meaningfully involved in the inventive process. Documenting the role of scientists, engineers, and data scientists in guiding AI systems can help establish human inventorship under existing patent laws.
Companies should also maintain detailed records of training data, model design, and human decision-making during the research process. Such documentation may prove crucial if inventorship is later challenged during patent prosecution or litigation.
For industries such as pharmaceuticals, AI-driven discovery is already accelerating innovation timelines. Studies have shown that patent data itself is becoming a key input for AI systems used in drug discovery. Protecting these innovations through carefully structured patent strategies will be essential for maintaining competitive advantage.
The Road Ahead: Defining the Inventor of the Future
The debate over AI inventorship ultimately reflects a deeper question about how legal systems adapt to technological change. Patent law has always evolved in response to new forms of innovation, but the rise of autonomous AI systems challenges some of its most fundamental assumptions. For India, the issue is not yet urgent in the courts, but it is rapidly approaching on the policy horizon. As AI-driven research becomes more common across sectors such as pharmaceuticals, climate technology, and advanced manufacturing, disputes over inventorship are likely to emerge.
The countries that respond proactively will shape the next generation of intellectual property frameworks. Those who delay may find themselves struggling to adapt to innovation ecosystems that no longer fit traditional legal categories. India, therefore, faces a strategic choice. It can maintain its current framework and rely on human inventorship doctrines, or it can begin developing policies that explicitly address AI-generated inventions. Either way, clarity will be essential.
The debate may not be about recognising machines as inventors, but about correctly attributing innovation in an AI-augmented world. Where AI systems operate on human-designed architectures, trained on human-created knowledge, and guided by human intent, the inventive essence may still lie with the human mind. For India, this perspective offers a pragmatic path forward: preserving the integrity of patent law while accommodating the transformative role of AI in modern innovation
The inventor of the future may still be human, but increasingly that human will be collaborating with machines capable of generating ideas at unprecedented speed. Patent systems must therefore evolve to recognise not only who invents, but how invention itself is changing. For innovators, lawyers, and policymakers alike, the challenge is clear: defining the legal boundaries of creativity in an age where intelligence is no longer exclusively human.
