AI Inventorship and the Emerging Global Consensus

AI Inventorship and the Emerging Global Consensus

This three-part Brooks Kushman series examines AI inventorship as a patent-law question with domestic, international, and portfolio-wide implications for companies integrating AI into research and development.

• Part 1 explains the U.S. baseline after Thaler v. Vidal and the USPTO’s November 2025 guidance on AI-assisted inventions.
• Part 2 compares the global treatment of AI inventorship, where courts and patent offices have largely converged on natural-person inventorship despite differences in reasoning and procedure.
• Part 3 turns that convergence into practical considerations for applicants managing AI-assisted innovation across international patent portfolios.

Artificial intelligence is now part of the inventive process in ways patent law must increasingly account for. AI systems may assist with searching technical space, generating candidate designs, modeling experimental outcomes, identifying patterns in data, or proposing solutions that human researchers later test and refine. For inventorship, those developments raise a focused question: can AI hold the legal status of “inventor”?

U.S. law gives a clear answer. The U.S. Court of Appeals for the Federal Circuit’s decision in Thaler v. Vidal held that an inventor under the Patent Act must be a natural person. [1] The United States Patent and Trademark Office’s November 2025 Revised Inventorship Guidance for AI-Assisted Inventions applies that rule in the more common setting of AI-assisted research, confirming that the use of AI does not create a separate inventorship standard. [2] AI may be a tool in the inventive process, but the legal inventor must be human.

Patent portfolios commonly span multiple jurisdictions, and inventor designation often travels with related filings, priority claims, assignments, declarations, and ownership narratives. In that setting, AI inventorship is not merely a local filing question. It asks whether the legal account of invention can remain coherent across systems that may differ in inventor-designation requirements, entitlement concepts, and procedures for correcting inventorship.

For international applicants, that reality frames the comparative question: is the American position exceptional, or does it reflect a wider global pattern? Surveyed decisions and office practice do not proceed along a single doctrinal path. Some authorities emphasize statutory text; others focus on legal capacity, entitlement, inventor-designation formalities, or the view that any expansion to machine inventors requires legislative action. Yet the result is increasingly consistent: across the jurisdictions surveyed, the dominant rule remains that inventorship is limited to natural persons.

–

Part 1.  AI Inventorship in the United States

1. The Historical Human Premise of U.S. Inventorship

American patent law begins from a constitutional premise that is broad in scope and human in orientation. Article I, Section 8 empowers Congress “[t]o promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries.” [3] This clause, referred to as the Intellectual Property Clause, does not define “inventor” in modern technical terms, and understandably did not contemplate artificial intelligence. The language, though, reflects a legal tradition in which invention was understood as an act of human originality, attribution, and contribution.

The Patent Act carries that premise forward. It defines an “inventor” as “the individual or, if a joint invention, the individuals collectively who invented or discovered the subject matter of the invention.” [4] In ordinary use, as the U.S. Supreme Court has recognized, the word “individual” points to a natural person. [5] The statutory structure reinforces the same assumption, including provisions requiring an inventor’s oath or declaration and using personal pronouns to describe the inventor’s belief that he or she is the original inventor. [6]

This human premise also appears in the doctrine of conception. U.S. inventorship has long focused not merely on who performed experiments, operated equipment, or reduced an idea to practice, but on who formed the inventive idea reflected in the claims. The traditional formulation defines conception as the formation, in the inventor’s mind, of a definite and permanent idea of the complete and operative invention. That language was developed long before modern AI systems, but it captures the same inherited assumption: inventorship is tied to a human act of mental conception.

AI inventorship now tests that inherited framework. The U.S. patent system was built around human acts of invention, while modern AI systems are alleged, at least in certain cases, to have produced inventive concepts without human conception.

1. Thaler v. Vidal: The Bright-Line Rule That AI Cannot Be an Inventor

The Federal Circuit confronted the AI-as-inventor question in Thaler v. Vidal, [7] a case involving two patent applications that named DABUS, an artificial intelligence system, as the sole inventor and identified no human inventor. Because the applications did not name a natural person, the USPTO had concluded that they lacked a valid inventor and were therefore incomplete. [8] Dr. Thaler challenged that decision, and the Federal Circuit affirmed. [9]

The court framed the issue succinctly: “This case presents the question of who, or what, can be an inventor.” [10] The court then narrowed the inquiry, explaining that the case did not require an abstract inquiry into machine creativity, machine rights, or the nature of invention itself. In the court’s words, “we do not need to ponder these metaphysical matters.” [11] The question instead began and ended with the statutory text.

The statutory analysis produced an unequivocal rule: “Here, there is no ambiguity: the Patent Act requires that inventors must be natural persons; that is, human beings.” [12]

1. USPTO Revised Guidance: AI Assistance Does Not Change the Inventorship Standard

In November 2025, the USPTO issued Revised Inventorship Guidance for AI-Assisted Inventions, rescinding and replacing its February 2024 guidance. [13] The revised guidance places AI-assisted inventions within ordinary inventorship doctrine. [14] As the USPTO states, “[t]he same legal standard for determining inventorship applies to all inventions, regardless of whether AI systems were used in the inventive process.” [15] The guidance states the point more directly: “There is no separate or modified standard for AI-assisted inventions.” [16]

The guidance also carries forward the rule from Thaler: AI systems “cannot be named as inventors or joint inventors on a patent application” because they “are not natural persons.” For purposes of U.S. inventorship, the relevant legal actor remains the natural person.

The guidance further explains that “AI systems, including generative AI and other computational models, are instruments used by human inventors.” It compares AI systems to laboratory equipment, computer software, research databases, and other tools that may assist in the inventive process. Just as a microscope, simulation platform, or research database does not become a co-inventor merely because it helped produce information, an AI system does not become a joint inventor merely because it generated outputs used by human researchers.

The revised guidance therefore keeps AI-assisted inventions within ordinary inventorship doctrine. It does not create a special AI-inventorship framework, and it does not treat AI assistance as a separate category of patentability. The analysis remains focused on the natural person or persons responsible for the claimed invention.

1. Where U.S. Law Stands: Human Conception Remains the Touchstone

Taken together, Thaler and the USPTO’s revised guidance establish two settled points and leave one practical inquiry. First, an AI system cannot be named as an inventor or joint inventor on a U.S. patent application. Second, the use of AI does not create a special inventorship standard or make an otherwise patentable invention unpatentable. The remaining question is the familiar, fact-intensive one: which natural person or persons conceived the claimed invention?

Thaler and the USPTO guidance address different postures. Thaler involved applications that named an AI system as the sole inventor and identified no human inventor. The USPTO’s revised guidance addresses the more common setting of human researchers using AI tools. In both settings, the analysis returns to human conception: whether one or more natural persons conceived the claimed invention under ordinary inventorship principles. As the USPTO states, “[t]he fact that AI tools were used in the development process does not change the joint inventorship analysis among the human contributors.” [17]

The revised guidance also narrows how the familiar joint-inventorship analysis fits into AI-assisted development. The USPTO withdrew its 2024 approach that had applied the Pannu factors to AI-assisted inventions generally, explaining that those factors apply when determining whether multiple natural persons qualify as joint inventors. Where one natural person used AI, the inquiry is whether that person conceived the invention under the traditional conception standard. Where multiple natural persons were involved, ordinary joint-inventorship principles apply among those human contributors. The application still may not list an AI system as inventor or joint inventor.

For U.S. applicants, the rule is clear but application will often be fact-specific. AI may assist, accelerate, or enrich the inventive process. It may generate ideas, outputs, candidate solutions, or experimental paths. The patent application must still identify the natural person or persons who conceived the claimed invention. In that sense, AI assistance does not create AI inventorship, and it does not displace the ordinary requirement that inventorship be traced to human conception.

The U.S. rule provides a clear baseline, but international patent portfolios rarely stop at one jurisdiction. Part 2 turns to the comparative question: whether the U.S. position is exceptional or part of a broader global pattern for applicants filing across jurisdictions.

AI Inventorship and the Emerging Global Consensus

This is Part 2 of a three-part Brooks Kushman series on AI inventorship. Part 1 addressed the U.S. baseline after Thaler v. Vidal and the USPTO’s November 2025 guidance. This part turns to the global landscape and asks whether the U.S. position is an outlier or part of a broader convergence on natural-person inventorship.

Part 2.  Global Convergence on Human Inventorship

The U.S. rule provides a clear starting point, but the comparative question is broader. Across the jurisdictions surveyed, inventorship remains limited to natural persons, and no examining patent regime surveyed has substantively recognized AI as an inventor. Figure 1 maps those categories, including the narrow set of AI-associated issued-right outcomes that arose through registration-based mechanisms.

Figure 1: Global AI Inventorship Positions. Plain labels identify jurisdictions that exclude AI as an inventor in the surveyed materials. Bracketed labels identify AI-associated issued rights arising through procedural or registration-based mechanisms rather than substantive recognition of AI inventorship.

The jurisdictions surveyed fall into two practical categories.

1) Jurisdictions excluding AI as an inventor:

• Australia (AU)
• Brazil (BR)
• Canada (CA)
• China (CN)
• European Patent Office / Europe (EP)
• Germany (DE) (patent)
• Israel (IL)
• Japan (JP)
• New Zealand (NZ)
• Republic of Korea (KR)
• Saudi Arabia (SA)
• Switzerland (CH)
• Taiwan (TW)
• United Kingdom (GB)
• United States (US)

2) Jurisdictions with AI-associated issued rights through registration-based mechanisms:

• Germany (DE) (GebrM utility model)
• South Africa (ZA)

The first category reflects the prevailing position across the surveyed examining regimes: an AI system cannot be named as inventor. The route to that result differs. Some authorities emphasize statutory text; others focus on legal capacity, entitlement, inventor-designation formalities, or institutional restraint in the face of technological change.

The European Patent Office (EPO) provides a useful starting point because its decisions make explicit how inventor designation functions under the European Patent Convention (EPC). The inventor designation is not merely a label; it is part of a legal structure that assumes a person capable of holding and transmitting rights.

1. European Patent Office: Inventor Designation and Legal Capacity

The EPO reached the natural-person rule through the formal requirements of the EPC. [18] Dr. Thaler’s European applications identified DABUS as inventor and asserted that he had acquired the right to the European patents as owner, employer, or successor in title to the AI system. The Receiving Section refused the applications, and the Legal Board of Appeal dismissed the appeal.

The Board’s decision states the rule plainly: “A machine is not an inventor within the meaning of the EPC.” [19] The reasoning joined inventor designation to legal capacity. The EPC required an inventor designation identifying a person, and a machine lacked legal personality to be an employee, transferor, or rights-holder. The EPO therefore supplies a clear “legal capacity” rationale for the human-inventor rule: whatever role an AI system plays in generating technical output, it cannot satisfy the inventor‑designation function the EPC assigns to a person capable of holding and transmitting rights.

1. United Kingdom: Statutory Inventorship and Entitlement

The United Kingdom reached the same result through the Patents Act 1977. Dr. Thaler filed two U.K. applications naming DABUS as the system that created the inventions and asserting entitlement because he owned DABUS. The UKIPO found that the applications failed to identify a person believed to be the inventor and failed to explain a legally sufficient derivation of title. The Supreme Court affirmed. [20]

The Court treated the issue as one of statutory interpretation. Under the Act, an “inventor” means the “actual deviser” of the invention. [21] The Supreme Court concluded that the statutory scheme “permit[s] only one interpretation”: an inventor “must be a natural person,” while DABUS was “not a person at all, let alone a natural person.” [22] Because DABUS could not be an inventor, Dr. Thaler also could not derive entitlement merely from ownership of the AI system.

The U.K. decision is particularly instructive on entitlement. It shows that, under the Patents Act 1977, ownership of an AI system does not substitute for the statutory route by which a non-inventor obtains rights from, or through, an inventor. The decision reaches the same bottom-line result, while adding a distinct entitlement-based explanation for why the application could not proceed.

1. Germany Patent Track: Natural-Person Inventorship with AI-Use Disclosure

Germany’s patent-track treatment confirms the natural-person rule while adding a narrower procedural refinement. Dr. Thaler’s German patent application originally identified the inventor as “DABUS,” with the statement that “[t]he invention was created independently by artificial intelligence.” [23] The German Patent and Trade Mark Office rejected the application on the ground that only a natural person could be named as inventor. In later proceedings, the Federal Patent Court set aside that rejection only as to Dr. Thaler’s third auxiliary inventor designation, which named “Stephen L. Thaler, PhD” and added that he “prompted the artificial intelligence DABUS to generate the invention.” The Federal Court of Justice dismissed both appeals, leaving that limited result in place.

The Federal Court of Justice stated the governing rule directly: “Only a natural person can be an inventor” under Section 37(1) of the German Patent Act, and a hardware or software machine system cannot be designated as inventor “even if it has artificial intelligence functions.” [24] The court also explained that the designation of a natural person remains “possible and necessary” when an AI system was used to find the claimed technical teaching. [25] In the court’s view, even a substantial AI contribution does not preclude identifying at least one natural person as inventor based on that person’s contribution.

The German refinement lies in how the inventor-designation record may describe AI’s role. The Federal Court of Justice held that Dr. Thaler’s third auxiliary designation was not a sufficient basis for refusing the application because it clearly named Dr. Thaler as inventor and made clear that DABUS was not being identified as a co-inventor, but only as “a means used by the applicant to find the claimed technical teaching.” [26] The additional AI-related statement was treated as legally irrelevant and separable from the inventor designation. Thus, Germany’s patent-track decision preserves the natural-person requirement while allowing, at least in that procedural posture, a limited acknowledgement that AI was used in arriving at the claimed teaching.

1. Australia: An Early Outlier Reversed on Appeal

Australia briefly appeared to be a substantive outlier, but the final appellate position aligns with the mainstream rule. The Deputy Commissioner of Patents initially determined that Dr. Thaler’s application did not comply with the Patent Regulations because it named DABUS, rather than a natural person, as inventor. On judicial review, the Federal Court set that decision aside, holding that an “inventor” under the Patents Act 1990 could be an artificial intelligence system or device.

That trial-level decision did not survive appeal. The Full Court of the Federal Court reversed, focusing on the relationship between inventorship and entitlement under the Australian patent statute. [27] The Court reasoned that the statutory scheme assumes an inventor from whom entitlement can flow, and that “the origin of entitlement to the grant of a patent lies in human endeavour.” [28] Having regard to the statutory language, structure, history, and policy of the Patents Act, the Full Court concluded that the application did not comply with the requirement to name an inventor by naming DABUS.

The Australia proceedings are notable because the mainstream position emerged only after appellate review. A trial court initially accepted AI inventorship, but the Full Court returned the analysis to the human premise of the Australian patent system: entitlement flows from an inventor, and the inventor contemplated by that framework is a natural person.

1. Japan: Statutory Interpretation and Legislative Restraint

Japan reaches the natural-person rule through statutory structure and institutional restraint. Dr. Thaler entered the Japanese national phase for the DABUS application and identified the inventor as “DABUS, the artificial intelligence that autonomously made the present invention.” The JPO ordered him to amend the national-phase document to state the name of a natural person as inventor. When he declined, the JPO dismissed the application. The Tokyo District Court upheld that dismissal, and the Intellectual Property High Court dismissed Dr. Thaler’s appeal. [29]

The Intellectual Property High Court treated the issue through the structure of Japan’s Patent Act. The court reasoned that the Act recognizes the right to obtain a patent, and provides a procedure for granting a patent, only with respect to inventions made by natural persons. It emphasized that the Act presupposes a natural-person inventor in application and publication procedures, including the requirement to state the inventor’s “name,” using terminology associated with the full name of a natural person. On that basis, the court concluded that “inventions” for which patents may be granted under the Act are limited to those for which natural persons are the inventors.

Japan therefore adds a legislative-restraint dimension to the global pattern. The court acknowledged that the Patent Act did not anticipate autonomous AI inventions, but treated that gap as a matter for legislative policy rather than judicial interpretation. In the court’s words, whether to grant patent rights for AI inventions requires legislative discussion based on the social and industrial-policy effects of such protection, and is “difficult to deal with” through interpretation of existing law. [30] On the current Japanese framework, the issue is not resolved by treating AI as the source of the inventive output; the patent system requires a natural-person inventor and leaves any broader protection for autonomous AI inventions to legislative design.

1. Canada: Natural-Person Inventorship Under the Patent Act and Patent Rules

Dr. Thaler’s Canadian application identified DABUS as the system that devised the claimed food container and enhanced-signaling device. The Patent Appeal Board reviewed whether the application complied with the Patent Act and Patent Rules, focusing on whether a valid “inventor” and, by extension, a valid applicant had been identified. The Commissioner of Patents refused the application in June 2025. [31]

The Canadian decision approached the issue through statutory interpretation. Although “inventor” is not defined in the Patent Act or Patent Rules, the Board concluded that the term is limited to a natural person or persons. It relied on the ordinary meaning of “inventor,” the statutory context linking inventorship to persons and legal representatives, and the structure of the patent system as one in which rights are awarded to, and transferred between, natural or legal persons. Because DABUS is an artificial intelligence system, it could not be recognized as the inventor, could not transfer rights, and could not support filing by an “inventor’s legal representative.” [32]

Canada adds a recent North American confirmation of a point also seen in the U.K. and EPO materials: the defect is not solved by ownership of the AI system or by characterizing the applicant as the AI’s legal representative. The Canadian framework still requires a valid inventor, and for that purpose the inventor must be a natural person.

1. Additional Jurisdictions Reflecting the Same Trend

New Zealand and Switzerland add useful nuance to the natural-person rule. In New Zealand, the High Court acknowledged that the Patents Act 2013 was not drafted in exactly the same terms as earlier New Zealand legislation and that the text of section 22 was “sufficiently wide” to make the AI-inventorship argument possible. [33] The court nevertheless found no indication that Parliament intended, in replacing the older “true and first inventor” language, to open the door to AI inventorship. The amendment instead addressed the narrower problem of excluding mere importers from inventor status. On that basis, the court declined to expand the statutory concept of inventor through interpretation, explaining that such a step was more properly reserved for Parliament. [34]

Switzerland reached the same broad result, but with a more AI-assisted-invention-oriented analysis. The Swiss Federal Administrative Court held that DABUS could not be registered as inventor and that a patent application could not proceed without naming an inventor. [35] At the same time, the court treated human involvement in the AI process as potentially relevant to the inventor-designation inquiry. It noted that a natural person may qualify through contributions to the AI data-processing process, and treated Dr. Thaler’s alleged involvement, including providing data, training DABUS, receiving the outputs, transmitting the invention to counsel, and recognizing patentability, as sufficient for the procedural issue before it. Switzerland therefore reinforces the natural-person rule while illustrating the harder follow-on question: which human involvement in an AI-assisted process is enough to support inventorship?

Taiwan provides another example of the mainstream rule, notable in that its reasoning links inventorship to both technical contribution and legal status. In the DABUS proceedings, TIPO required correction of the application to identify a natural person as inventor; when the applicant maintained that DABUS was the sole inventor, the application was not accepted, and the courts affirmed. [36] The Supreme Administrative Court stated that an inventor must be a person who actually contributes to the technical features of the claimed invention and must be a natural person.

In China, current CNIPA examination guidance states that the inventor must be a natural person and that AI cannot be listed as inventor. [37] Brazil, Saudi Arabia, and Korea point in the same direction. Brazil’s patent office treated inventorship as limited to natural persons; Saudi Arabia rejected the DABUS application on the ground that an AI system is not a natural person with legal capacity; and Korea likewise rejected DABUS as inventor. Taken together, these shorter-form authorities point to the same practical baseline: across these systems, an AI system may not be placed on the inventor line.

1. Procedural Counterpoints: Issued Rights Without Substantive Recognition

The principal examining regimes reflect the dominant rule. Two AI-associated issued-right outcomes, however, require separate consideration: South Africa’s patent registration and Germany’s utility model registration.

Both identified an AI system on the inventor line, but neither supplies a reasoned holding that AI may be an inventor under a substantive inventorship standard. South Africa’s outcome arose in a registration-based patent system, and Germany’s utility model registration is distinct from Germany’s patent-track rule, under which the Federal Court of Justice required a natural-person inventor. These counterpoints illustrate how procedural design can yield AI-associated issuance without implying doctrinal recognition of AI inventorship in an examining regime.

This global convergence has practical implications for international applicants. Part 3 addresses inventor naming, human-contribution records, and portfolio consistency for AI-assisted inventions.

AI Inventorship and the Emerging Global Consensus

This is Part 3 of a three-part Brooks Kushman series on AI inventorship. Part 1 addressed the U.S. rule after Thaler v. Vidal and the USPTO’s November 2025 guidance. Part 2 surveyed the global convergence on natural-person inventorship. This final part turns to practical implications for applicants managing AI-assisted innovation across international patent portfolios, including inventor naming, invention-disclosure practices, and consistency across related filings.

Part 3.  Portfolio Implications for AI-Assisted Inventions

1. Treat AI Inventorship as a Portfolio-Consistency Issue

For international applicants, AI inventorship is best understood as a portfolio-consistency issue rather than a jurisdiction-shopping opportunity. Across major examining regimes, the operative baseline remains natural-person inventorship. Meaningful AI involvement in research and development, standing alone, does not support placing the AI system on the inventor line.

Inventor designation also carries consequences beyond the application form. In many systems, it bears on entitlement, assignment, declarations, ownership, publication, and later validity challenges. A position taken in one jurisdiction may create friction in another, especially where related applications share a common priority claim or a common account of how the invention was made. A coherent portfolio strategy should begin with the human contributors, then describe AI’s role with precision.

1. Build an Invention Record Around Human Conception

When AI tools are used during development, the record should remain focused on the claims and the human contribution to them. In-house invention disclosure forms can help by including a short AI-use prompt asking whether AI tools were used, what outputs were generated, and which human contributors selected, modified, validated, or translated those outputs into claimed subject matter. The point is not to label every AI interaction as inventive, but to preserve the facts counsel will need to evaluate conception and inventorship.

Those facts do not automatically make every participant an inventor. Inventorship remains a legal conclusion tied to the claims and governed by the law of the relevant jurisdiction. Some applicants may prefer to avoid creating unnecessary AI-use narratives and rely on inventor declarations and ordinary invention records unless a real AI-inventorship issue is present. That approach reduces over-documentation risk, but it can leave counsel with a thin record if inventorship is later questioned. A balanced record should identify the human acts that shaped, selected, and completed the inventive concept without overstating AI’s legal role.

1. Distinguish AI Assistance from AI Inventorship

AI use and AI inventorship should remain separate concepts. Patent systems are not generally rejecting inventions because AI tools were used. The legal issue arises when an application identifies no human inventor and seeks to name the AI system itself as inventor.

Drafting and prosecution should reflect that difference. Depending on the facts, AI may accurately be described as a tool, system, model, or source of outputs. Identifying AI as the legal inventor carries a different meaning, and current authorities overwhelmingly reject that position. The global pattern does not place AI-assisted inventions outside the patent system; it keeps the legal inventor human.

1. Watch for Reform, but Draft Under Current Law

Many courts and patent offices recognize that AI-generated invention raises serious policy questions. Those questions include whether patent incentives are needed to encourage development of inventive AI systems, whether disclosure will be lost to trade secrecy if AI-generated outputs cannot be protected, who should own rights in machine-generated inventions, and whether a new statutory or sui generis regime would be preferable to adapting existing patent law.

For now, those questions have generally been treated as matters for legislatures, patent offices, or international policy processes, not as grounds for judicial recognition of AI inventorship. Current filing practice should proceed under the law as it stands: identify the human inventor or inventors, document the human contribution, and describe AI’s role carefully without allowing technological novelty to obscure the legal requirement of human inventorship.

Takeaway

U.S. law provides a clear baseline, and the global authorities point in the same direction: AI may assist in developing new technology, but current patent systems still require the legal account of inventorship to run through human contributors.

For international applicants, the practical lesson is precision. As AI becomes more deeply embedded in research and development, the most durable filing strategy will be one that keeps the human contribution in view. That means identifying the natural person or persons responsible for the claimed invention, preserving the record of how human contributors shaped the inventive concept, and describing AI’s role without overstating its legal status.

[1] Thaler v. Vidal, 43 F.4th 1207 (Fed. Cir. 2022), cert. denied, 143 S. Ct. 1783 (2023).

[2] Revised Inventorship Guidance for AI-Assisted Inventions, 90 Fed. Reg 54636 (Nov. 28, 2025), available at https://www.federalregister.gov/d/2025-21457.

[3] U.S. Const. art. I, § 8, cl. 8.

[4] 35 U.S.C. § 100(f).

[5] See Mohamad v. Palestinian Auth., 566 U.S. 449, 454 (2012).

[6] 35 U.S.C. § 115(b)(2).

[7] Thaler v. Vidal, 43 F.4th 1207.

[8] Id. at 1208.

[9] Id. at 1213.

[10] Id. at 1208.

[11] Id.

[12] Id. at 1210.

[13] Revised Inventorship Guidance for AI-Assisted Inventions, supra note 2.

[14] An “AI-assisted” invention refers to an invention in which one or more human beings used AI as part of the inventive process. An “AI-generated” invention refers to an invention alleged to have been conceived autonomously by an AI system, with no human inventor identified. The former asks how ordinary human inventorship principles apply when AI was used as a tool; the latter asks whether AI itself can be the inventor.

[15] Id.

[16] Id.

[17] Id.

[18] ECLI:EP:BA:2021:J000820.20211221 (Legal Bd. App.), decision dated July 5, 2022, available at https://www.epo.org/en/boards-of-appeal/decisions/j200008eu1.

[19] Id.

[20] Thaler v Comptroller-General of Patents, Designs and Trade Marks, [2023] UKSC 49; decision dated December 20, 2023; available at https://supremecourt.uk/uploads/uksc_2021_0201_judgment_3f445a5dc7.pdf.

[21] Id. at ¶ 26.

[22] Id. at ¶ 56.

[23] German Patent and Trade Mark Office v. Thaler, X ZB 5/22; dated June 11, 2024; available at https://www.bundesgerichtshof.de/SharedDocs/Entscheidungen/DE/Zivilsenate/X_ZS/2022/X_ZB___5-22.pdf?__blob=publicationFile&v=1.

[24] Id. at ¶ a).

[25] Id. at ¶ b).

[26] Id. at ¶ 64.

[27] Comm’r of Patents v. Thaler, [2022] FCAFC 62 (Austl. Full Fed. Ct. Apr. 13, 2022), available at https://www.judgments.fedcourt.gov.au/judgments/Judgments/fca/full/2022/2022fcafc0062.

[28] Id. at ¶ 116.

[29] Intell. Prop. High Ct. (Japan), Reiwa 6 (Gyo-Ko) No. 10006 (Jan. 30, 2025), English translation available at https://www.courts.go.jp/ip/eng/assets/ip/eng/chizai_en/chizai_en-pdf-3624.pdf.

[30] Id. at 9.

[31] Thaler, Stephen L. (Re), 2025 CACP 8 (Can.), available at https://brevets-patents.ic.gc.ca/opic-cipo/cpd/eng/patent/3137161/summary.html.

[32] Id. at ¶ 79.

[33] Thaler v Commissioner of Patents [2023] NZHC 554 (Mar. 17, 2023), available at https://www.justice.govt.nz/jdo_documents/workspace___SpacesStore_03467327_0e3d_41de_b88e_da0dd54b8116.pdf.

[34] Id. at ¶ 33.

[35] Fed. Admin. Ct. (Switz.), Ruling B-2532/2024 (June 26, 2025), available at https://www.bvger.ch/media-releases/41f1e066-84f9-4492-a7bb-b4941dc02790/en/b-2532_2024_web.pdf.

[36] Thaler v. Taiwan Intell. Prop. Off. (TIPO), 110 Xing Zhuan Su 3 (IPC Ct. Aug. 2021).

[37] China Nat’l Intell. Prop. Admin. (CNIPA), webpage available at https://www.cnipa.gov.cn/art/2024/12/31/art_66_196988.html.