A US District Court judge recently answered this question in the negative. Indeed, the court stated that the “clear answer is no.” See Thaler v. Hirshfeld, Case No. 1:20-cv-903 (US Dist. E.D. Virginia September 3, 2021). The court held that the “plain language” of the US Patent Act provides that only inventors are entitled to patents and that only natural persons — human beings — can be inventors. See media report here.

The case involved two inventions that were purportedly solely created by an artificial intelligence (“AI”) machine. The owner — Stephen Thaler — gave his invention-creating AI machine the name of DABUS. When Thaler filed patent applications with the US Patent & Trademark Office (“USPTO”), he listed DABUS as the “inventor” and asserted that DABUS had “assigned” to Thaler its rights in the inventions allowing Thaler to file the applications on its behalf. The USPTO rejected the patent applications asserting that only human beings can be “inventors.” Without a human inventor signing the applications, the USPTO refused to process the applications. Eventually, Thaler filed suit in federal court seeking to overturn the USPTO’s refusal to process the applications.

But, as noted, the US federal court agreed with the USPTO. The court’s decision was based on the repeated use of the word “individual” in the Patent Act. The court held that the ordinary meaning of the word “individual” refers to a “natural person” which, in legal terms, means a human being.

Thaler faced a similar result in the United Kingdom. See BBC media report here. Thaler filed patent applications with the UK Intellectual Property Office which rejected the applications for the reasons as cited by the USPTO. On September 23, 2021, a UK Court of Appeals (by a 2-1 vote) affirmed the UK IPO’s decision. As reported, the majority wrote: “Only a person can have rights. A machine cannot. A patent is a statutory right and it can only be granted to a person.”

However, interestingly enough, Thaler has found some success in Australia. See Guardian media report here. On behalf of DABUS, Thaler has filed for patents in more than a dozen countries. His Australian patent applications were rejected by the Australian Patent Office. But, in July 2021, an Australian federal court overruled the rejection and ordered the Australian Patent Office to process the applications. The court argued that allowing machine AIs to own patents was consistent with the underlying purposes of patent law — to encourage and foster innovation.

Here in the United States, likely, if AI machines are going to be allowed to file patent applications, the Patent Act will have to be changed.

For more information or if you are a human and have an invention or design that you want to patent, contact the patent lawyers at Revision Legal at 231-714-0100.

The USPTO’s Current Position on AI Inventorship

Since the Thaler v. Hirshfeld decision, the USPTO has issued formal guidance reinforcing its position that AI cannot be listed as an inventor on a patent application. In February 2023, the USPTO published guidance stating that only natural persons qualify as inventors under 35 U.S.C. § 100(f), which defines “inventor” as “the individual or, if a joint invention, the individuals collectively who invented or discovered the subject matter of the invention.” The agency specifically interpreted “individual” to require a natural person, consistent with the Supreme Court’s interpretation of that term in Mohamad v. Palestinian Authority, 566 U.S. 449 (2012).

What this means practically is that any patent application that lists an AI system as the sole inventor will be rejected during examination. Applications naming a human being as a co-inventor alongside an AI may proceed, but only if that human made a genuine inventive contribution to the claimed invention. Token or nominal human involvement will not suffice. Patent examiners are trained to scrutinize inventorship declarations and the scope of claimed inventions in AI-assisted cases.

What Happens When AI Substantially Contributes to an Invention?

The more difficult — and practically important — question is how to handle inventions where AI played a substantial role in conception but one or more humans also contributed. Under 35 U.S.C. § 116, all persons who contributed to the conception of at least one claimed invention must be named as inventors. “Conception” is the touchstone of inventorship: it is defined as “the formation in the mind of the inventor of a definite and permanent idea of the complete and operative invention.” Burroughs Wellcome Co. v. Barr Labs., Inc., 40 F.3d 1223 (Fed. Cir. 1994).

When an AI system generates a novel output — a drug candidate, a circuit design, a chemical formula — and a human engineer then selects, refines, and claims that output, the human’s contribution to “conception” is contested terrain. The USPTO’s guidance suggests that humans who direct, select, and apply AI-generated outputs may qualify as inventors, while those who merely follow instructions from an AI without exercising independent judgment likely do not. This is an area of law that will be intensely litigated in the coming years.

The Federal Circuit’s Affirmance and Its Implications

The Fourth Circuit’s decision in Thaler v. Vidal, 43 F.4th 1207 (Fed. Cir. 2022), affirmed the district court ruling. The Federal Circuit held that under the Patent Act, inventors must be natural persons. The court acknowledged that Congress could change the law to permit AI inventorship but held that absent such a change, it was bound by the statute’s plain text. The court declined to address broader policy arguments about whether AI inventorship would promote innovation, viewing those as legislative rather than judicial questions.

Thaler’s petition for certiorari to the Supreme Court was denied in April 2023, leaving the Federal Circuit’s decision as binding precedent. This means that for now, the rule in the United States is clear: AI cannot be an inventor. What remains unclear is how courts will evaluate inventorship in cases where AI and humans collaborate closely — and whether Congress will eventually act.

Strategies for Protecting AI-Assisted Inventions

Despite the prohibition on AI inventorship, businesses can and should take affirmative steps to protect innovations that were developed with significant AI assistance. Here are the primary strategies:

• Document human inventive contribution carefully — Maintain detailed records showing the specific contributions each human inventor made to the conception of the claimed invention. Lab notebooks, commit histories, design documents, and email chains can all serve as evidence of human inventive contribution.
• Work with patent counsel from the outset — When AI is involved in the inventive process, engaging a patent attorney early allows for strategic drafting of claims that emphasize human conception and control. Claims can be crafted to cover the human-directed aspects of the invention rather than the raw AI output.
• Consider trade secret protection — For AI-generated innovations that cannot easily be patented or where inventorship is unclear, trade secret law may offer an alternative form of protection. Unlike patents, trade secrets do not require disclosure and do not require inventorship declarations.
• Monitor international developments — The global landscape is in flux. South Africa issued the first patent listing an AI as inventor in 2021. Australia’s initial pro-AI decision was later overturned on appeal. The European Patent Office, like the USPTO, requires human inventors. Businesses operating internationally should monitor these developments and develop jurisdiction-specific IP strategies.

The Legislative Outlook

Congress has begun to take notice of the AI inventorship question. Several proposals have been floated to amend the Patent Act to address AI-assisted and AI-generated inventions. One approach would define a new category of “AI-assisted invention” with distinct ownership rules directing patent rights to the humans or entities that deploy the AI. Another approach would allow an AI system’s owner or operator to be designated as the legal inventor in limited circumstances. None of these proposals has yet been enacted into law, but the pace of legislative interest is accelerating alongside the rapid deployment of large language models and generative AI tools in research and development contexts.

For businesses investing heavily in AI-driven R&D, this legislative uncertainty creates real IP risk. A patent obtained today based on AI-assisted invention could face invalidity challenges if the underlying law changes or if courts develop new standards for evaluating human contribution. Working with experienced patent counsel to build a robust documentation record is essential.

If you have questions about patenting an AI-assisted invention or about inventorship requirements under current USPTO rules, contact the patent lawyers at Revision Legal at 231-714-0100.