Revised Inventorship Guidance for AI-Assisted Inventions
Authored by: Jeremy J. Gustrowsky
The United States Patent and Trademark Office (USPTO) recently released revised guidance clarifying how inventorship should be determined for inventions created with the help of artificial intelligence (AI). This new guidance, issued in November 2025, rescinds the earlier “Inventorship Guidance for AI-Assisted Inventions” from February 2024 and replaces it with a more straightforward approach. The updated policy is a response to the rapidly growing use of AI tools in research and development, and it aims to answer a pressing question: What is the proper legal standard for determining inventorship when AI is involved?
In short, the new guidance states that only natural persons—meaning human beings—can be named as inventors on U.S. patent applications, even if AI played a significant role in the inventive process. The Federal Circuit has firmly established that AI systems, no matter how sophisticated, cannot be listed as inventors or joint inventors because they are not “natural persons.” This means that, according to the USPTO, AI is treated as a tool, much like a microscope or a computer program, rather than as a co-inventor.
The USPTO’s new guidance emphasizes that the traditional legal standard for inventorship applies equally to inventions developed with AI assistance. The central issue is “conception,” which is defined as the formation in the mind of the inventor of a definite and permanent idea of the complete and operative invention. To be considered an inventor, a person must have a clear mental picture of the invention and be able to describe it in detail, such that someone with ordinary skill could make and use it without undue experimentation.
Importantly, the guidance clarifies that the mere use of AI tools does not automatically make the user an inventor. The human involved must have contributed to the conception of the invention. If multiple people are involved, the traditional rules for joint inventorship apply. Each person must have made a significant contribution to the conception or reduction to practice of the invention, beyond simply explaining well-known concepts or using the AI as a black box.
The guidance also addresses the situation where an application lists an AI system or other non-human entity as an inventor. In such cases, the USPTO will reject the application for failing to comply with the legal requirements for inventorship. Only humans can be named, and applications must be corrected to remove any non-natural persons from the inventor list.
Notably, the revised guidance extends to all types of patents, not just utility patents. This includes design and plant patents. For example, if a new plant variety is bred with the help of AI, the human who contributed to the creation and recognized the uniqueness of the plant is the inventor, not the AI system that may have suggested genetic combinations or analyzed data.
The USPTO also clarified how inventorship rules apply to patent applications that claim priority to earlier U.S. or foreign applications. Both the earlier and later applications must name the same human inventor or at least one joint inventor in common. If a foreign application lists an AI as an inventor, a U.S. application claiming priority to it must only list the human inventor(s) to comply with U.S. law.
In summary, the USPTO’s revised guidance provides an important update on the evolving standards at the USPTO for inventors, companies, and patent practitioners navigating the intersection of AI and patent law. This latest message clarifies a departure from the human vs AI team-oriented approach of the past opting for a more “AI is just a tool” approach. It reinforces the point of view that only humans can be recognized as inventors under U.S. patent law. We can expect the USPTO to continue adjusting its perspective as it seeks to maintain rules for inventorship grounded in existing established legal principles, even as the capabilities of AI continue to expand.