The accelerating capabilities of artificial intelligence (AI), namely autonomous and generative models’ ability to reason and create, have brought into focus a foundational question in intellectual property law: can a non-human entity, such as an AI system that independently conceives, generates, and validates a patentable invention, be legally recognized as its inventor?
This question challenges not only statutory frameworks and doctrinal interpretations of inventorship across leading jurisdictions, but also broader policy considerations concerning innovation incentives, legal coherence, and the attribution of responsibility. Answering this question requires an examination of how major legal systems currently define inventorship, how courts interpret these provisions in the context of AI-generated inventions, and what potential pathways exist to reconcile this emergent technology with enduring legal principles.
Case Law in Major Jurisdictions
A number of recent legal cases held in the US, Australia, and the UK involving patents for AI-generated inventions highlight the subject of this article. Examining the rulings in these cases will establish views on AI inventorship in the eyes of the law in their respective countries. Disseminating how each juristiction delt with attribution of AI-generated inventions will serve as the starting point for a rich discussion on this topic.
United States: Thaler v. Vidal, Federal Circuit, 2022
In Thaler v. Vidal, 43 F.4th 1207 (Fed. Cir. 2022), Dr. Stephen Thaler attempted to obtain U.S. patents naming his AI system, called DABUS, as the sole inventor. He asserted that there was no human contribution to the “conception” of the inventions under consideration. The USPTO denied the applications, the Eastern District of Virginia affirmed, and the Federal Circuit confirmed that under U.S. statute, an inventor must be a natural person. Central to that result is 35 U.S.C. § 100(f), which defines “an ‘inventor’ as ‘the individual or, if a joint invention, the individuals collectively who invented or discovered the subject matter of the invention.’” Because “individual,” in federal statutory interpretation, ordinarily denotes a human being unless Congress clearly indicates otherwise, the Court held that an AI system cannot be an inventor under current U.S. law, determining“that only a natural person can be an inventor, so AI cannot be.”
The ruling further emphasizes that statutory language such as “individual” within the Patent Act is not explicitly defined to encompass non-human entities. Additionally, other legal requirements, such as the necessity for an inventor’s oath or declaration, implicitly assume human attributes, including the capacity for belief and the ability to provide a signature.
Australia: Commissioner of Patents v Thaler [2022] FCAFC 62
Australia presents a useful contrast. In a first instance decision in 2021 (Thaler v. Commissioner of Patents [2021] FCA 879), the Federal Court (Beach J) held that under the Patents Act 1990 (Cth) an AI system could be designated as inventor for formal purposes, given that the Act does not define “inventor,” and that the term “person” in s.15 includes legal persons under the Acts Interpretation Act; but ownership had to lie with a natural or legal person. That decision was reversed by the Full Court in Commissioner of Patents v. Thaler [2022] FCAFC 62, which unanimously rejected the view that a non‑human can be inventor. The Full Court held that in s.15(1), the inventor named in a patent specification must be a natural person; even the alternative limbs for entitlement under sections (b), (c), (d) presuppose that the inventor in s.15(1)(a) is human; title or assignment must trace from a human inventor.
The Full Court concluded that the Deputy Commissioner had properly determined the application to be non-compliant with regulation 3.2C(2)(aa). This conclusion was grounded in a holistic reading of the Patents Act, its text, internal logic, legislative history, and the broader policy aims it seeks to advance, all of which point to the requirement that an inventor must be a natural person. The naming of DABUS, a non-human entity, was therefore found to be legally insufficient.
United Kingdom: Thaler v. Comptroller‑General of Patents, Designs and Trade Marks [2023] UKSC 49
In the United Kingdom, DABUS cases culminated in the Supreme Court judgment on 20 December 2023, Thaler v. Comptroller‑General [2023] UKSC 49. The facts were parallel: Dr. Thaler filed patent applications under the UK Patents Act 1977, naming DABUS as inventor. The United Kingdom Intellectual Property Office (UKIPO) refused them for failure to name a human inventor pursuant to sections 7 and 13, and Dr. Thaler appealed. The Supreme Court unanimously held that under the 1977 Patent Act, an “inventor” must be a natural person. DABUS is not a “person, let alone a natural person,” and thus not an inventor.
The Court further rejected the argument that ownership of AI (i.e. Dr. Thaler’s ownership in DABUS) gives entitlement under section 7(2)(b) or (c) of the 1977 Act, or to apply for a patent by derivation or succession. That carefully structured code in section 7 requires an inventor (a person) and then entitlement that flows from that inventor; ownership of a machine itself does not suffice. Because no person was identified as inventor, the patent applications were declared withdrawn (statutorily required under section 13 when inventor is not identified).
Doctrinal Foundations and Interpretive Principles
From the leading U.S., Australian, and U.K. decisions, certain legal, interpretive principles emerge, which jointly explain why courts have so far refused to recognize AI as inventor under existing law.
- Statutory Definition and Language: The definition of “inventor” in the U.S. law (35 U.S.C. § 100(f)) requires an “individual … who invented or discovered”; in the UK law (Patents Act 1977) section 7 and 13 require naming of inventor(s); in Australia, s.15 requires the inventor be a “person.” The choice of words such as “individual,” “deviser,” and “person” are interpreted according to ordinary meaning, legislative context, and precedents. Courts insist upon natural person status where “person” or “individual” appear without qualification (e.g., Thaler v. Vidal in U.S.; Commissioner of Patents v. Thaler in Australia; Thaler v. Comptroller‑General UKSC).
- Interpretive Presumptions: Courts rely on the presumption that statutory terms referring to persons or individuals refer to humans unless otherwise specified. This is reinforced by precedent such as Mohamad v. Palestinian Authority, 566 U.S. 449, 454 (2012), where “individual” was held to mean a natural person. In Thaler v. Vidal, the Federal Circuit cited Mohamad to affirm that “individual” does not include AI absent clear legislative signal.
- Statutory Structure and Coherence: Identification of the inventor is not mere formality; many doctrines (entitlement, ownership, assignment, duties of inventor, oath/declaration) are tied to the inventor’s identity and legal capacity. If the inventor is non‑human, many legal consequences (e.g., signing an oath, being subject to false marking, ownership, assignable rights) become incoherent. Courts have observed that statutory schemes assume human capacity to undertake legal acts.
- Historical and Legislative History: Judges invoke historical foundations (e.g., Statute of Monopolies 1623 in Australia, earlier British statutes) and previous laws (earlier Patents Acts) in interpreting terms. In Australia, pre‑1952 Acts, legislative history, and common law understanding of “inventor” have always presupposed human agency. Legislatures have not amended their IP statutes to permit non‑human inventors, though the topic has been raised.
- Limitation by Legislative Silence: Because of these established presumptions and doctrines, the absence of explicit statutory language authorizing non‑human inventor status is taken as determinative. In the United States and the United Kingdom, there has been no amendment or legislative history that clearly signals an intention to recognize AI as inventor. Courts have refused to expand such definitions by implication.
Policy and Innovation Trade‑offs
Although prevailing legal frameworks deny the possibility of recognizing AI systems as inventors, this position is not without significant costs. Excluding AI from inventorship may undermine innovation policy, distort the incentives for disclosure, and create legal uncertainty in cases where human contribution is minimal or ambiguous. As such, any serious proposal for legislative or doctrinal reform must carefully balance these trade-offs, weighing the legal, ethical, and technological implications of extending or withholding formal recognition of AI-generated contributions within the patent system.
Incentives and Innovation
If AI‑generated inventions cannot be patented because no human inventor can be identified, this might disincentivize investment in the generation of autonomous inventions. Innovators may try to manufacture or exaggerate human contribution to satisfy legal requirements, potentially compromising the integrity of the inventive process. Alternatively, inventions may be kept secret or placed in trade‑secret protection if patent protection is unavailable, reducing public disclosure which remains one of the core rationales of patent law.
Legal Certainty
Recognition of AI as an inventor invites significant legal complexity by raising a number of difficult questions: What counts as “conception” by a machine? How does one assess an inventive step or non‑obviousness when the means of invention is non‑human? How to assign or derive ownership? Who bears liabilities for inventorship mis‐declaration or false statement? How priority (the right of an applicant to claim an earlier filing date based on a prior application for the same invention) or derivation (a situation where one party claims to be the inventor but has actually taken the invention or idea from someone else, without being the true originator) work in patent law when the chain of human agency is partial or missing?
Administrative and Examination Burdens
Patent offices already wrestle with identifying human inventors in complex collaboration settings. Incorporating AI as inventor would require new guidelines, perhaps evidence of AI logs, provenance, autonomy, dataset usage, and perhaps audits. Examiners would need to assess whether AI truly “devised” or “conceived” novelty. Offices would potentially see a higher volume of applications with ambiguous inventorship.
Philosophical and Ethical Considerations
Some may object that attributing inventorship to machines undermines the notion of human agency, rewards, or responsibility. Inventorship carries not only economic but also moral and reputational dimensions. Further, the legal system traditionally links patent awards to individuals who can contribute, understand, and are accountable. Recognizing non‑human inventorship may blur the lines of responsibility, especially in misuses, harms, or invalid patents.
What the Current Laws Permit
Existing laws support human contribution and AI as a tool. Currently, an invention that is partially or predominantly facilitated by AI is not ruled out. What matters is whether a naturalperson makes a significant contribution to conception or “devising” of at least one claimed element of the invention. That is to ensure at least one human being meets the statutory inventorship criteria. The U.S. Federal Circuit, in Thaler v. Vidal, and Australian Full Court, in Thaler (2022), both implicitly accept that AI used as a tool by humans is compatible with patentable inventions, provided the human meets the inventorship threshold. Ownership, assignment, and entitlement issues then flow from those named human inventors.
Table 1: Comparative Summary; Do Current Statutes Allow for AI Inventorship?
| Jurisdiction | Statute/Key Provision | Inventorship Requirement | Recognizes AI as Inventor? | Key Reasoning for Exclusion |
|---|---|---|---|---|
| USA | 35 U.S.C. § 100(f), §100(g) etc. | “Inventor” must be “the individual…who invented or discovered.” | No | “Individual” means a natural person; inventor oath/declaration presumes human; no statutory signal to the contrary. Thaler v. Vidal. |
| Australia | Patents Act 1990 (Cth) s.15(1) + Regulations | Inventor named must be a “person” but judgment clarified “inventor” must be a natural person. | No (as of 2022) | Although “person” includes legal entities, inventor in s.15(1)(a) must be natural; historical meaning; entitlement structure requires relation to human inventor. Commissioner of Patents v Thaler [2022] FCAFC 62. |
| UK | Patents Act 1977, sections 7, 13 | Inventor must be “actual deviser,” a natural person under statutory language or code. | No | Comptroller‑General [2023] held statute requires human; ownership of AI does not suffice; inventor identification requirement under section 13 mandates human inventor. |
What Reforms Would Be Needed to Permit AI Inventorship
To permit AI inventorship under current legal systems, substantial reforms, both legislative and doctrinal, would be required across multiple dimensions of patent law. These reforms would need to address not only statutory language and interpretive doctrines, but also the underlying purposes of the patent system, including the roles of attribution, accountability, and incentive structures. Listed below are the types of reform that would be necessary to grant legal recognition of AI systems as inventors:
Statutory Redefinition of “Inventor”
Patent statutes would have to be amended to include non-human entities, such as software agents or autonomous AI systems, within the definition of “inventor.” In such a scenario, legislatures would introduce specific provisions recognizing “machine-originated” inventions, provided certain conditions are met, or revise existing language to refer to “agents” or “entities” without limiting inventorship to natural persons.
Clarifying Ownership of AI-Generated Inventions
Even if AI is named as an inventor, it cannot own property or assert legal rights, at least in the U.S., Australia, and the UK. To resolve this, relevant laws would need to designate who owns an invention made by AI (e.g., AI’s developer, operator, or owner). To help ensure patent rights are properly transferred and enforceable, these designations would have to be incorporated into national and international IP statutes, with clear allocation mechanisms.
Reforming the Inventor’s Oath and Declaration Requirements
Current patent rules assume inventors can sign documents and attest to the originality of their invention. The reforms would allow a human proxy, such as the owner of the AI, to sign on AI’s behalf or introduce exceptions to these formalities for AI-generated inventions. This would likely require a blend of administrative rulemaking and legislative amendments.
Revisiting Inventive Step or Nonobviousness Standards
Current patent laws require inventions to be nonobvious to a “person skilled in the art.” The principle of nonobviousness is used to determine if an invention is a sufficiently inventive step beyond what is already publicly known. If AI routinely produces results beyond human expectations, it may challenge how “inventiveness” is assessed. In permiting AI inventorship, legal systems would need to clarify that AI-generated inventions are still evaluated from a human perspective, or else risk raising the standard for all inventors, human and non-human alike. Alternatively, there could be a dual threshold standard (one for human inventors and another for AI) or specific recognition that AI may assist in invention without automatically rendering the outcome obvious. This reform is more doctrinal than statutory but would require judicial reinterpretation or guidance from patent offices.
Enhancing Disclosure for AI-Generated Inventions
To ensure transparency and accountability in AI-generated inventions, patent applicants should be required to disclose detailed information about the AI’s role in the invention process. This includes the specific AI systems and models used, the training data and tunable hyperparameters, the model inputs or prompts used in generating the invention, or the degree of human intervention. By mandating such disclosures, the patent system can better adapt to the evolving nature of invention without sacrificing its core principles of clarity, accountability, and public trust. It ensures that as we integrate AI into the realm of innovation, we do so with the transparency necessary to maintain a fair and functional intellectual property system.
Reforming International Agreements and Harmonization
Global IP systems are governed by treaties such as the Paris Convention and the TRIPS Agreement, none of which contemplate non-human inventors. “The TRIPS Agreement, which came into effect on 1 January 1995, is to date the most comprehensive multilateral agreement on intellectual property.” Unilateral changes by a single country could introduce legal asymmetries or undermine reciprocity. A reform in this regard would necessitate pursuing international harmonization by amending treaties (a lengthy and politically challenging process), issuing interpretative declarations or soft law instruments recognizing AI inventorship as a prerogative, and creating bilateral or regional agreements to test new recognition frameworks.
Balancing Innovation and Policy Risks
Recognizing AI as an inventor could have unintended policy consequences, such as undermining human-centered innovation incentives, eroding responsibility, or triggering a flood of algorithmically generated patents. Any reform in this area must be carefully designed to include safeguards such as requiring human oversight or attribution to remain central in the patent process and limiting overuse by large entities. Such reforms may not be purely legal in nature, but rather involve regulatory policy and ethics governance frameworks, potentially informed by public consultation.
Legal and Societal Considerations
AI inventorship has sparked intense debate across legal and societal contexts, raising fundamental questions about the invention process and protections as well as inventor responsibilities and ownership. AI inventorship also raises broader concerns about fairness, transparency, accountability, and equitable access to the benefits of AI-driven innovation. Among these, the economic impact, particularly the role of patent protection in incentivizing investment and research, represents a key aspect. This section explores the various interconnected challenges and considers how legal systems might adapt to the evolving realities of invention in the age of machine intelligence.
Legal Questions
The prospect of recognizing AI systems as inventors under patent law engenders a host of intricate legal questions. These questions interrogate the fundamental premises of existing intellectual property frameworks and challenge longstanding doctrinal assumptions:
- Threshold of Autonomy. To what extent must the inventive act, such as conception or formulation of an inventive concept, be the product of autonomous machine activity to qualify for inventor status? Specifically, how much independence must an AI system exhibit before it can be recognized as an inventor? If human involvement includes designing, training, supplying data, or setting objectives, does that suffice to attribute inventorship to the human, or should the AI itself be credited as the inventor?
- Attribution of Inventive Step and Novelty. How should credit and responsibility be allocated when the inventive step emerges from autonomous or probabilistic processes (e.g., agentic AI or stochastic modeling)? AI that use these processes are set apart in that they can omit human decisioning. What methodologies apply in assessing prior art and nonobviousness for inventions generated without direct human conception?
- Ownership, Entitlement, and Assignability. Assuming AI is recognized as the inventor, who holds legal ownership of the patent? Since ownership must vest in a natural or juridical person, does it derive from the AI’s developer, operator, programmer, or financer? How do principles of entitlement and derivation apply in such contexts, particularly in the absence of clear statutory guidance?
- Liability and Legal Obligations. Inventorship imposes legal duties, including the obligation to disclose relevant information, certify the accuracy of inventor declarations, and avoid false statements. Given that AI systems lack legal personhood, can they bear these obligations, or must responsibility be allocated to a human being or corporate entity?
- Proof and Evidentiary Standards. If AI inventor status is to be recognized, then what types of evidence should patent offices require to substantiate an AI system’s role as inventor? How can documentation such as activity logs, provenance records, and training datasets be effectively utilized in evaluating AI inventorship? What methodologies can distinguish between human and AI-generated contributions to the inventive process? What standards of proof are appropriate for verifying AI’s inventive involvement in patent applications? How can patent offices balance the need for transparency with the protection of trade secrets and proprietary information during examination of AI-generated inventions?
Arguments in Favor of the Legal Recognition of AI as Inventors
- Promoting Innovation and Technological Progress. Recognizing AI as an inventor aligns patent law with the new role technology plays in the inventive process and encourages development and deployment of advanced AI systems. If AI-generated inventions are systematically excluded from patent protection, innovation incentives may fade and in turn discourage new investments into AI research and development. Granting inventor status to AI would better reflect and stimulate the evolving landscape of creativity.
- Fair Attribution of Inventive Credit. From an ethical standpoint, attributing inventorship to the correct entity respects the factual origin of inventions. If the creative process is effectively autonomous, then excluding AI from inventor recognition misattributes credit to human actors who may have had only ancillary involvement. Upholding principles of honesty and fairness in the patent systemrequires that the true author of an invention is acknowledged, even in the case of AI.
- Maintaining Legal and Doctrinal Coherence. The patent system’s purpose is to reward inventive activity that advances useful art. Practically, this reward should be tied to the actual inventor. Policy reform to allow AI inventorship would restore consistency of doctrine and mitigate legal uncertainty.
- Encouraging Transparent Disclosure and Accountability. Recognizing AI as an inventor may incentivize clearer documentation of AI involvement and promote transparency in patent applications. This aligns with the goals of the patent system to disseminate knowledge and enable technological progress. By obligating applicants to explicitly address the role of AI, such reform would enhance the quality of patent disclosures.
- Supporting Equitable Access to Intellectual Property Rights. If AI systems are becoming key drivers of innovation, failing to recognize them as inventors risks consolidating control of patent rights in the hands of those who merely own or operate AI systems, rather than those whose creative input genuinely shapes the invention. Reform introducing normative safeguards to balance rights among developers, users, and financers would promote fairness in IP ownership distribution.
- Reflecting and Upholding Societal Values on Creativity and Agency. Granting inventor status to AI challenges anthropocentric views but can be justified by evolving societal understandings of creativity as a process not exclusively bound to human cognition. Adapting patent law accordingly respects pluralistic conceptions of inventive agency consistent with contemporary technological capabilities.
Arguments Against the Legal Recognition of AI as Inventors
- Preserving the Human-Centric Foundation of Patent Law. Patent systems have traditionally been designed to reward human ingenuity, reflecting deeply held values about creativity, moral agency, and personal contribution. Extending inventorship to AI risks undermining this foundational principle, potentially eroding the human-centric nature of intellectual property rights that reinforce accountability and ethical responsibility.
- Concentration of Patent Power and Economic Inequality. A key concern is that recognizing AI as inventors may disproportionately benefit large corporations with substantial resources to develop and deploy advanced AI systems, thereby exacerbating existing economic disparities. This could marginalize individual inventors and smaller entities by concentrating patent ownership and innovation opportunities into the hands of economically powerful players, undermining fairness and equitable access within the patent system.
- Protecting Legal Clarity and Doctrinal Stability. Recognizing AI as an inventor introduces complex legal uncertainties and may disrupt established patent doctrines, including notions of authorship, ownership, and liability. Preserving doctrinal clarity and predictability supports legal certainty, which is crucial for maintaining trust in the patent system and avoiding inadvertent harm to innovation ecosystems.
- Avoiding Unintended Consequences and Systemic Abuse. Granting AI inventor status could incentivize opportunistic behavior, such as mass patent filings by AI with minimal human oversight, overwhelming patent offices and diluting patent quality. The principle of a balanced patent system that promotes genuine innovation over procedural exploitation opposes such reform, so long as safeguards against this kind of exploitation are absent.
- Upholding Accountability and Responsibility. Inventorship carries legal and ethical responsibilities such as ensuring accuracy of disclosures and adherence to ethical standards. Since AI lacks moral agency and cannot bear liability, recognizing it as an inventor may undermine mechanisms for accountability, a pillar of the patent system in preserving integrity.
- Preserving Incentives for Human Creativity and Investment. There is a concern that recognizing AI inventors could diminish incentives for human creators and investors who drive technological advancement. Rewarding AI directly might redirect resources away from human innovation or disrupt existing incentive structures designed to nurture human inventive activity.
- Maintaining Societal and Ethical Norms Regarding Creativity and Personhood. Granting legal inventorship to AI may challenge societal and ethical norms that associate creativity with personhood, consciousness, and intentionality. This may raise profound questions about the moral and legal recognition of non-human entities, potentially upsetting broader legal and societal frameworks.
Conclusion
The question of whether AI should be officially recognized as an inventor within patent law presents a multifaceted challenge that intersects legal doctrine, technological capability, and policy considerations. Current statutory frameworks and prevailing judicial interpretations remain anchored to a human-centric concept of inventorship, reflecting deep-seated legal traditions and practical requirements of accountability, creativity, and ownership. However, as the inventive capabilities of AI systems improve, lawmakers will face mounting challenges in ignoring the legal strain introduced by this technology. Arguments in favor of reform emphasize alignment with innovation incentives, accurate attribution, and doctrinal coherence, while equally weighty counterarguments underscore the risks of legal uncertainty, erosion of human creativity incentives, and potential socioeconomic disparities.
Moving forward, any reform must carefully balance these competing interests, ensuring that patent law remains both effective and just in incentivizing genuine innovation, whether human, artificial, or a hybrid thereof. The evolving technological landscape demands a nuanced, interdisciplinary approach that thoughtfully reconciles legal principles with the transformative potential of AI, thereby safeguarding the integrity and purpose of the patent system.
Given the complexity and novelty of this subject, further empirical and doctrinal research is necessary. Through comprehensive, forward-looking research and dialogue, the legal community can better anticipate and shape the role of AI in the inventive process, ensuring that intellectual property law evolves in step with transformative technologies while upholding its core mission to foster innovation for the public good.
Future studies should explore how different jurisdictions might harmonize their patent frameworks to address AI inventorship, including comparative analyses of statutory interpretations and case law developments. Additionally, interdisciplinary collaboration between legal scholars, technologists, ethicists, and economists can develop robust frameworks that balance innovation incentives with societal values and equity concerns. Empirical research into the real-world impacts of AI-generated inventions on innovation ecosystems, patent quality, and market dynamics will also provide critical insights to inform policy decisions. Finally, exploring the potential for novel legal categories or sui generis rights tailored to AI-generated innovations may offer a constructive path forward that accommodates technological advances without compromising foundational legal principles.
ABOUT ENTEFY
Entefy is an enterprise AI software company. Entefy’s patented, multisensory AI technology delivers on the promise of the intelligent enterprise, at unprecedented speed and scale.
Entefy products and services help organizations transform their legacy systems and business processes—everything from knowledge management to workflows, supply chain logistics, cybersecurity, data privacy, customer engagement, quality assurance, forecasting, and more. Entefy’s customers vary in size from SMEs to large global public companies across multiple industries including financial services, healthcare, retail, and manufacturing.
To leap ahead and future proof your business with Entefy’s breakthrough AI technologies, visit www.entefy.com or contact us at contact@entefy.com.