How Global Patent Law Universally Rejects AI Inventors Over the Definition of "Conception"

On August 27, 2024, the German Federal Court of Justice established a definitive legal boundary for artificial intelligence, ruling that an AI system cannot be legally recognized as the inventor on a patent application. The decision marked the culmination of a multi-year, 17-jurisdiction legal campaign designed to force courts to answer a fundamental question: who owns an idea when a machine does the thinking?

By rejecting the premise of machine inventorship, the court reinforced a centuries-old legal doctrine that intellectual property rights are designed exclusively to incentivize and reward human cognitive effort, regardless of how sophisticated autonomous systems become.[4]

The German ruling aligned Europe’s largest economy with the United States and the United Kingdom, cementing a global consensus that intellectual property rights are exclusively reserved for human beings.

However, the German court introduced a critical nuance that separates it from its international peers: while the legal title of "inventor" must belong to a natural person, the AI system can be explicitly named in the patent specification to transparently document how the novel design was generated. This creates a hybrid framework where legal ownership remains human, but technical attribution accurately reflects the machine's role.[2]

This legal architecture addresses a growing reality in corporate research and development departments worldwide. From pharmaceutical laboratories discovering novel protein structures to engineering firms optimizing complex semiconductor layouts, AI systems are increasingly performing the cognitive labor traditionally classified as "conception." When a machine generates a viable product design without direct human prompting, the existing legal frameworks struggle to map 20th-century statutory definitions onto 21st-century autonomous capabilities, forcing patent offices to untangle exactly where the human contribution ends and the machine's work begins.[3]

The test case that forced this global judicial reckoning centers on DABUS, which stands for Device for the Autonomous Bootstrapping of Unified Sentience, an artificial intelligence system created by computer scientist Stephen Thaler. Unlike standard machine learning models that are trained to solve specific, narrowly defined problems, DABUS was engineered with two distinct neural networks: one designed to generate novel ideas by combining disparate concepts, and a second network designed to evaluate those generated ideas for real-world utility and novelty.[2]

In 2018, the DABUS system autonomously generated two distinct, highly specific inventions without direct human instruction: a fractal-based food container designed to change shape for easier stacking and improved heat transfer, and a neural flame device intended to attract attention during search-and-rescue operations. Thaler filed patent applications for both of these inventions across 17 different global patent offices, deliberately listing the DABUS system as the sole inventor while listing himself as the legal patent owner by virtue of owning the machine.[1]

The filing strategy was explicitly designed by Thaler and his legal team to test the boundaries of international statutory law. By refusing to list a human inventor on the paperwork, Thaler forced national patent offices to make a binary choice: either grant property rights to a machine, or reject novel, demonstrably useful inventions entirely because of their non-human origin. The United States Patent and Trademark Office formally rejected the applications in 2020, prompting a high-stakes federal lawsuit that would eventually reach the appellate courts.[1]

The legal barrier in the United States is deeply rooted in the specific statutory language of the 1952 Patent Act, which governs all intellectual property claims in the country. Under 35 U.S.C. Section 100(f), an inventor is strictly defined as the "individual or, if a joint invention, the individuals collectively who invented or discovered the subject matter of the invention." The legal debate hinged entirely on whether the word "individual" could be interpreted broadly enough to encompass an autonomous software system.[1][2]

In August 2022, the United States Court of Appeals for the Federal Circuit issued a binding ruling in the case of Thaler v. Vidal, firmly shutting the door on AI inventorship. Writing for the unanimous three-judge panel, Judge Leonard Stark stated the legal reality plainly and without caveat: "Here, there is no ambiguity: the Patent Act requires that inventors must be natural persons; that is, human beings." The court relied heavily on Supreme Court precedent establishing that statutory references to individuals inherently mean human beings.[1][2]

The Federal Circuit determined that because the artificial intelligence is not a natural person, it is legally incapable of forming the "mental conception" required by United States patent law. Conception is legally defined as the formation in the mind of the inventor of a definite and permanent idea of the complete and operative invention. Because a machine does not possess a legally recognized "mind," it cannot conceive of an invention, rendering the DABUS applications fundamentally invalid under current federal statutes.[1]

The United Kingdom reached an identical legal conclusion in December 2023, further isolating the DABUS campaign. The UK Supreme Court unanimously dismissed Thaler’s final appeal, ruling that under the Patents Act 1977, an inventor must be a natural person. The court concluded that the UK Intellectual Property Office was therefore entirely justified in withdrawing the patent applications when Thaler repeatedly refused to name a human inventor, establishing a firm precedent that UK law will not accommodate machine-generated intellectual property without an act of Parliament.[2][4]

This unified global rejection creates a significant and growing policy friction for the technology sector. The World Intellectual Property Organization has explicitly noted that the core purpose of the international patent system is to incentivize innovation by granting a temporary commercial monopoly in exchange for public disclosure of the technology. If AI-generated inventions cannot be patented, companies may rationally choose to keep their breakthroughs hidden as proprietary trade secrets, fundamentally undermining the foundational goal of the patent system and slowing the broader pace of scientific advancement.[3]

Furthermore, the strict requirement to name a human inventor forces modern technology companies into a legally precarious position. If a pharmaceutical company utilizes an advanced AI model to discover a new drug compound, but lists a human operator as the inventor merely to satisfy the USPTO's paperwork requirements, they risk having the patent invalidated years later during litigation. Competitors could successfully argue fraudulent inventorship if the named human's actual contribution is deemed legally insufficient to qualify as true conception.[2]

This is precisely where the August 2024 German Federal Court of Justice ruling offers a potential, albeit limited, compromise for the industry. By strictly requiring a human being to be listed as the legal inventor on the primary application, but explicitly allowing the artificial intelligence system to be documented in the technical specification, the German legal system effectively separates legal ownership from technical attribution. This dual-track approach acknowledges the reality of machine generation while maintaining the human-centric focus of property law.[4]

This hybrid approach allows companies operating in Germany to secure vital intellectual property rights without committing perjury regarding the true origin of the invention.

It provides a formal mechanism for transparency, ensuring that the public scientific record accurately reflects the role of artificial intelligence in the creative process, while giving corporate investors the legal certainty they need to fund expensive research and development operations powered by autonomous systems. Legal scholars suggest this model could serve as a blueprint for other European nations struggling to balance innovation incentives with traditional legal doctrines.[3]

However, the German compromise does not resolve the underlying tension in major jurisdictions like the United States, where the legal standard remains rigid. The USPTO issued updated guidance in early 2024 stating that AI-assisted inventions are indeed patentable, provided a human being made a "significant contribution" to the actual conception of the invention. This guidance attempts to thread the needle between human and machine, but it leaves the exact threshold of human involvement dangerously ambiguous for patent attorneys and their corporate clients.[1][2]

The definition of a "significant contribution" remains highly subjective and entirely untested in federal court. Merely prompting an AI system with a general problem is legally insufficient for inventorship, but training a proprietary model on a highly specific dataset to solve a targeted engineering challenge might cross the threshold. Until the courts establish clear, bright-line rules for human-AI collaboration, companies are operating in a legal gray area, forced to guess how much human oversight is required to secure a patent.[2]

Until legislative bodies step in to formally update the statutory definitions of conception and inventorship, the legal framework will continue to lag significantly behind the technical reality of modern research. The DABUS test cases have definitively proven that while machines possess the technical capacity to invent novel products, the law remains an exclusively human domain. The next frontier of intellectual property law will not be about whether AI can invent, but how much human involvement is required to claim the machine's work as our own.[4]