The Next Tech Frontier: AI, Inventorship and Patents

The global patent system was designed during the Industrial Revolution to reward human ingenuity and foster technological progress. But today, we find ourselves in the midst of an AI revolution, where algorithms are no longer just tools for calculation – they are actively designing new chemical compounds, optimizing manufacturing hardware, and engineering novel mechanical structures.

This shift has created a profound tension within intellectual property systems worldwide. When an algorithm behaves like an inventor, how do patent offices respond? The ongoing debate centers on three complex areas: patentable subject matter and the non-human inventor, infringement, and enforcement.

The Non-Human Inventor: Can a Machine Hold a Patent?

“[The Congress shall have Power … ] To 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.”

U.S. Constitution, Art. I, Sec. 8, Cl. 8

The most fundamental issue in patent law is the question of inventorship. Traditionally, patent systems have operated under the assumption that an inventor must be a natural person.

The legal landscape has been heavily shaped by global test cases – most notably those involving DABUS (a creative AI developed by Stephen Thaler). Thaler filed patent applications worldwide listing DABUS as the sole inventor of a food container and an emergency beacon.

The response has been overwhelmingly clear in that an AI machine cannot be named as an inventor. Courts in the United States, the European Patent Office, and the United Kingdom have uniformly ruled that relevant patent statutes explicitly require an inventor to be a human being. Patent offices, such as the USPTO, have clarified that while AI tools can be used in the inventive process, a patent will only be granted if a human made a significant contribution to the conception of the invention.

This scenario creates a high stakes legal tightrope. If an AI autonomously discovers a groundbreaking new pharmaceutical molecule with minimal human intervention, that discovery might fail to meet patentable subject matter requirements entirely, leaving massive corporate investments unprotected.

Infringement: Who is Liable When an Algorithm Copies?

Patent infringement occurs when an entity makes, uses, sells, or imports a patented invention without authorization. In the context of AI, tracking and proving infringement introduces a unique set of headaches on both the input and output sides.

Upstream Software Infringement

To build a powerful AI system capable of automated design, developers write complex machine learning models. If these models utilize patented software architectures, neural network structures, or data processing techniques without a license, the developers are committing direct patent infringement.

The Downstream Black Box Problem

The trickier dilemma happens when an AI model independently generates a product or process that happens to duplicate an existing patent. If an autonomous manufacturing AI alters a factory blueprint to maximize efficiency, and that new blueprint inadvertently mirrors a competitor’s patented design, who is liable?

Because neural networks operate as a black box – where the exact decision-making process is obscured within millions of mathematical parameters – proving how the AI arrived at the infringing output, and assigning intent or direct liability to the human operator, is incredibly difficult under current legal frameworks.

Enforcement: Sifting Through an Automated Avalanche

Even if infringement can be theorized, executing enforcement in an AI-driven market is an uphill battle for patent holders.

Overwhelming the Patent Offices

The sheer velocity at which AI can iterate designs means that human engineers can suddenly file thousands of micro-variations of a single technology. This threatens to flood patent offices with low-quality, marginally distinct applications. This dynamic strains the examination process and makes it harder for legitimate patent holders to police their space against overlapping claims.

Rethinking POSITA

In patent litigation, courts judge whether an invention is obvious by looking through the lens of a hypothetical person of ordinary skill in the art (POSITA). If a standard engineer could have easily thought of the invention, it cannot be patented. As AI tools become standard equipment for every engineer, the baseline capabilities of a POSITA are shifting dramatically. If a basic AI program can automatically generate hundreds of optimal solutions to a mechanical problem in minutes, then what qualifies as truly nonobvious? This shifting baseline could create a scenario where enforcing existing patents and defending new ones become an ever-moving target.

Looking Forward

The cross-section of artificial intelligence and patent law has forced a re-evaluation of the very nature of human innovation. While patent registries are holding the line on human-centric inventorship, the sheer pace of AI development will continue to create tensions with the human inventorship requirement.

This blog posting is for informational purposes only. If you have a specific issue or question related to patents or AI, please contact Yonaxis I.P. Law Group.