The U.S. Court of Appeals for the Federal Circuit recently confirmed in Thaler v. Vidal that artificial intelligence (AI) agents cannot be listed as an inventor on a patent because the plain text of the Patent Act requires that inventors must be human beings. Patent law jurisprudence around the world has traditionally recognized only humans as inventors to patent applications, and this ruling likely concludes the American front of Stephen Thaler and the Artificial Intelligence Project’s years-long, global campaign advocating for patent laws to recognize AI agents as patent inventors.
As part of its concerted effort to overturn this convention, Thaler’s group submitted patent applications listing an AI agent named “DABUS” as the sole inventor to various patent offices. Nearly every jurisdiction, including Australia, the European Patent Office (EPO), Germany, and the United Kingdom rejected DABUS applications for failing to list a human inventor, confirming that each jurisdiction’s patent law only recognizes humans as inventors. The sole exception was South Africa, but that country lacks a substantive examination process.
Background: USPTO and Federal District Courts Reject AI-Inventorship
In the United States, a DABUS patent application was rejected during the pre-examination phase, when the U.S. Patent and Trademark Office (USPTO) reviews certain formal requirements of the new application, such as confirming whether the inventors are properly listed. The USPTO subsequently issued a Notice to File Missing Parts for failing to identify an inventor. Thaler petitioned for a review of the deficiency, claiming that inventorship should be extended to AI agents and as such DABUS was properly listed as the inventor for the application.
The USPTO’s disagreed and its Office of Petitions issued a decision explaining the USPTO’s position that an AI agent cannot be listed under U.S. patent law. In the decision, the USPTO pointed out that statutory language of various portions of the Patent Act (35 U.S.C. § 100 et seq.) suggest that an inventor must be human. Thaler challenged this decision first to the U.S. District Court for the Eastern District of Virginia, but the District Court agreed with the USPTO’s conclusions and rejected his arguments. Thaler then proceeded to appeal to the U.S. Court of Appeals for the Federal Circuit.
Statutory Language Recognizes Only Human Inventors, According to the Federal Circuit
The Federal Circuit agreed with the District Court and USPTO that the statutory language of the U.S. Patent Act only recognizes human inventors. “[T]here is no ambiguity: the Patent Act requires that inventors must be natural persons; that is, human beings.”
The Federal Circuit explained the Patent Act defines an “inventor” and “co-inventor” as an “individual.” A “person” could be a corporation or other entity, so there must be a narrower term. The term “individual,” according to the Supreme Court, “ordinarily mean[s] a human being, a person” distinguishable from artificial entities.
The Federal Circuit also noted other signals that Congress intended only human beings to be those “individuals” who qualify as “inventors.” The Act requires inventors to perform certain actions only performed by a human, such as executing an oath: “individual believes himself or herself to be the original inventor… of a claimed invention in the application.” In addition, “the Act uses personal pronouns – ‘himself’ and ‘herself’ – to refer to an ‘individual,’” but it “does not also use ‘itself,’” signaling to the court that Congress never intended to permit non-human inventors.
The Federal Circuit’s statutory approach in this case departs from earlier decisions at the USPTO, District Court, and foreign tribunals. Those earlier decisions focused their analysis on the definition of inventor in terms of “conception.” Each decision arrived at the conclusion that only the human mind conceives of an invention. Here, the Federal Circuit’s analysis is more fundamental and straightforward — an AI agent is not an “individual” eligible to be an “inventor” — and does not even engage in the discussion of “conception.”
Stephen Thaler’s Arguments
The Federal Circuit did not find any of Thaler’s points, referring to several portions of the Patent Act in support of his position that the “inventor” should be read broadly to include AI agents, as persuasive and upheld the USPTO’s and the District Court’s holdings.
Thaler first pointed to the use of “whoever” in the definition of the conditions for an inventor in 35 U.S.C. § 101: “[w]hoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title”, and the same term in relation to non-human entities in the context of infringement in 35 U.S.C. § 271:“in setting out what constitutes infringement, repeatedly uses ‘whoever’ to include corporations and other non-human entities.”
The Federal Circuit, however, was not persuaded by Thaler’s interpretation of the term “whoever.” Section 101 requires the putative inventor (i.e., “whoever”) to satisfy the “conditions and requirements,” which includes the definition of “inventor.” An AI agent would fail this condition. Section 271 describes who infringes a patent and defines the infringing acts, but it does not relate to who qualifies as an inventor. For example, nobody disputes that a non-human corporation can infringe a patent, but cannot be the inventor of a patent.
Thaler also argued that AI agents must qualify as inventors because patentability of AI-generated inventions would hinge on “the manner in which the invention was made” in violation of 35 U.S.C. § 103. The Federal Circuit similarly highlighted that § 103 is “not about inventorship.” This provision of § 103 precludes prejudice against inventions based on how an invention is made, but § 103 “does not trump a provision that specifically addresses who may be inventor.”
In addition, Thaler invoked the US Constitution’s Copyright and Patent Clause, arguing that allowing AI agents to be listed as inventors would also further “promote the Progress of Science and useful Arts.” According to Thaler, “inventions generated by AI should be patentable in order to encourage innovation and public disclosure.”
The Federal Circuit found this too speculative and does not weigh favorably against the plain text of the Patent Act. The Constitutional clause raised by Thaler grants Congress the legislative power to enact laws for copyright and patents. The Patent Act, adopted by Congress within its Constitutional jurisdiction, does not recognize AI agents as inventors and “Thaler does not (and cannot) argue that limiting inventorship to human beings is unconstitutional.”
The Federal Circuit also alluded to the question of AI-as-an-inventor versus AI-as-a-tool for invention: “whether inventions made by human beings with the assistance of AI are eligible for patent protection.” The Court left the question unanswered, leaving open the possibility that an invention made by a human using AI is patent eligible. Prior commentary indicates a consensus among the patent community that under patent law humans that use an AI agent as a tool are deemed inventors as long as the humans contributed to the conception.
It remains to be seen whether there will be any legislative impetus to expand inventorship to beyond natural human inventors. For now and the foreseeable future, under U.S. patent law all patent applications must only list human inventors, even if the involved individuals believe that use of an AI model was instrumental in the development of the invention.