Swiss Court Signals More Lenient Approach to AI Inventions

On June 26, 2025, the Swiss Federal Administrative Court (“Court”) issued its decision in case B-2532/2024, resolving a high-profile dispute over whether an artificial intelligence (“AI”) system can be named as an inventor under Swiss patent law. The case arose from a patent application for a food container that credited DABUS, an AI system developed by Dr. Stephen Thaler, as the sole-inventor.¹ The Swiss Federal Institute of Intellectual Property (“IPI”) had rejected the application because it listed only DABUS as inventor, reasoning that the Swiss Patents Act limits inventorship to natural persons.² Thaler requested that the Court (1) uphold the original AI inventorship designation; (2) grant the application without a named inventor; (3) name himself as the inventor; or (4) remand to the IPI for reconsideration.

The Court rejected Thaler’s first request, and affirmed the IPI’s position agreeing that, under the Swiss Patent Act, “inventor” refers to a human being who has contributed through intellectual creativity, making the naming of at least one natural person a mandatory element of a valid filing.³ The Court noted that current AI systems still rely on human input such as data preparation and recognition of the invention, i.e., identifying the AI output as patentable.⁴

In addressing Thaler’s second request, the Court issued a much more consequential ruling. In examining the record, it found that Thaler had been involved in data provision, training DABUS, and recognizing the DABUS’s output as a patentable invention.⁵ This combination of activities, in the Court’s view, satisfied the legal requirement for inventorship.⁶ As made clear in an English press release a week following the decision, “[p]atent applications must name a natural person as the inventor. A person who contributes substantially to the AI data treatment process, recognises its outcome as a patentable invention, and applies for patent protection also qualifies as an inventor.”⁷

Crucially, the Court’s ruling ordered the IPI to proceed with examination based on a corrected inventor designation naming Thaler as the applicant.⁸ In accepting Thaler’s second request, the Court signaled to applicants that AI-assisted inventions can still be patented in Switzerland if human contributions are properly documented and credited. In so doing, Switzerland aligns with the European Patent Office, the United States, and the United Kingdom, all of which have refused to accept DABUS as an inventor while leaving the door open for patents that list qualifying human inventors.⁹

The Swiss approach reinforces the growing global consensus that AI-assisted inventions remain protectable when supported by verifiable human contributions. While national laws vary in their specific statutory language and administrative guidance, the prevailing direction among major patent offices is to reject AI-only inventorship while preserving protection for inventions where a natural person played a demonstrable role in conception or recognition.¹⁰ This alignment, and the Swiss articulation of sufficient human involvement, offers valuable guidance to internationally operating patentees and their businesses.

Companies operating on the forefront of technological development can embrace the advent and growing sophistication of artificial intelligence if they ensure that their AI-assisted inventions rigorously and transparently reflect adequate human contribution. In this respect, Switzerland may have set a relatively low standard of contribution; recall, human recognition that the AI-output is patentable can clear the Swiss hurdle. The United States Patent and Trademark Office (“USPTO”), for example, requires significant human contribution for patentability. While a subjective threshold, USPTO guidance has suggested that the more documentation of human involvement, the better; extensive prompting of the AI model that ‘created’ an invention can be evidence of significant human contribution.¹¹

In highly competitive and heavily litigated fields such as pharmaceuticals, this guidance becomes even more crucial. AI tools are now frequently used from early drug development onward, including in molecular design, pharmacodynamic behavior prediction, drug repurposing, and even clinical trial recruitment.¹² To comply, companies should embed detailed disclosure practices into their innovation protocols, capturing when and how individuals interact with AI systems and ensuring that inventor declarations truthfully reflect those contributions.¹³ In United States patent litigation, including Hatch-Waxman disputes, inventorship has long been a potential vector for invalidity or unenforceability challenges under 35 U.S.C. § 102(f) prior to the Leahy-Smith America Invents Act, which prohibited a person from obtaining a patent if “he did not himself invent the subject matter sought to be patented.”¹⁴ Post-AIA, inventorship related disputes are likely to arise under 35 U.S.C. § 115, which requires a proper inventor declaration.¹⁵ Regardless, companies can help protect the patentability of their drugs by affirmatively involving personnel in each step of an inventive process touched by AI, thereby limiting the opportunity for challengers to assert a lack of human inventorship.

The Court’s decision underscores that, while AI may be a transformative tool in the inventive process, patent systems remain anchored to the principle of human inventorship. By clarifying that even modest but identifiable human contributions can render an invention patentable, the ruling underscores a growing consensus on the importance of accurate documentation and attribution of human input at every stage of the inventive process. For companies in sectors where patent rights are often the subject of intense litigation, these practices will be essential to securing and maintaining enforceable protection. As AI becomes more integrated into standard R&D workflows, jurisdictions that adopt clear, practical thresholds for human involvement are likely to influence the evolution of both patent law and AI-driven research.