USPTO Revises Guidance on AI-Assisted Inventorship
Artificial intelligence (“AI”) has rapidly become an integral part of research and development across nearly every industry. Engineers use AI to optimize product designs. Pharmaceutical companies employ AI to identify potential drug candidates. Software developers rely on generative AI to accelerate coding and problem-solving. As AI tools become increasingly sophisticated, an important legal question has emerged related to patents: Who is the inventor when AI contributes to an invention?
The United States Patent and Trademark Office (USPTO) addressed this issue in February 2024 by issuing guidance on AI-assisted inventorship. In November 2025, the agency issued revised guidance refining its earlier AI-specific framework and further aligning it with longstanding principles of U.S. patent law. The updated guidance withdraws certain AI-specific examples and reaffirms that traditional inventorship doctrine continues to govern AI-assisted inventions.
The revised guidance reinforces an important point: artificial intelligence may assist with the inventive process, but only natural persons may be named as inventors on a U.S. patent application. Rather than creating new legal standards specifically for AI-generated inventions, the USPTO emphasizes that established inventorship principles continue to apply regardless of whether AI played a role in developing the invention.
For businesses, inventors, universities, startups, and patent practitioners, the revised guidance provides additional clarity while underscoring the importance of carefully documenting each human inventor’s contribution. Companies increasingly incorporating AI into research and development should understand how these guidelines affect patent strategy, inventorship determinations, and patent application preparation.
Understanding Inventorship Under U.S. Patent Law
Inventorship is not simply a question of who worked on a project or who owned the technology. Instead, it is concerned with who conceived the claimed invention. Proper inventorship is critical because an incorrectly identified inventor may jeopardize the validity or enforceability of a patent if not properly corrected. Several provisions of the Patent Act govern inventorship, including:
35 U.S.C. § 100(f), which defines an inventor (or, in the case of joint inventions, inventors) as the individual(s) who invented or discovered the subject matter of the invention.
35 U.S.C. § 101, which provides that a patent may be obtained by a person who invents or discovers patent-eligible subject matter.
35 U.S.C. § 115, which requires inventors to execute an oath or declaration.
35 U.S.C. § 116, which governs joint inventorship and recognizes that multiple individuals may contribute to conception of a claimed invention.
Federal courts have long held that inventorship depends upon conception of the claimed invention. Conception generally refers to the formation, in the inventor’s mind, of a definite and permanent idea of the complete and operative invention as it will ultimately be claimed.
Importantly, not everyone who contributes to a project qualifies as an inventor. For example, individuals who merely:
Follow instructions,
Perform routine testing or experimentation,
Build prototypes without contributing to conception,
Reduce an invention to practice without contributing to the inventive concept, or
Provide general research or technical support
are generally not inventors unless they also contributed to the conception of at least one claimed invention (i.e., at least one of the claims in the patent). These longstanding principles remain unchanged by the USPTO’s revised AI-assisted inventorship guidance.
Why the USPTO Revised Its AI Inventorship Guidance
When the USPTO first issued AI inventorship guidance in February 2024, artificial intelligence technologies (particularly generative AI) were evolving at an extraordinary pace. The agency sought to provide practical examples illustrating how existing inventorship principles could be applied when AI systems participated in the inventive process.
The November 2025 guidance reflects a refinement of that earlier approach. Rather than continuing to rely on AI-specific examples that could be read as suggesting a distinct analytical framework, the USPTO reaffirmed that existing patent law already provides the appropriate standards for determining inventorship.
Accordingly, the revised guidance withdraws certain AI-specific examples and emphasizes that examiners and applicants should apply traditional inventorship principles (i.e., focusing on conception) regardless of whether AI tools were used during research or development. In other words, AI-assisted inventions are not subject to a separate body of inventorship law. They are evaluated under the same legal standards that have long governed inventorship determinations under the Patent Act and Federal Circuit precedent.
This clarification streamlines the analysis and reinforces the USPTO’s position that artificial intelligence functions as a tool in the inventive process, while inventorship remains grounded in human conception.
The Core Principle: Only Natural Persons May Be Inventors
The revised guidance reiterates one of the most important principles in U.S. patent law: only natural persons may qualify as inventors under the Patent Act.
This principle has been firmly established in federal case law, most notably Thaler v. Vidal, in which the United States Court of Appeals for the Federal Circuit held that the Patent Act limits inventorship to natural persons. As a result:
AI systems cannot be listed as inventors.
Machine learning models cannot sign or execute inventor declarations or legal instruments.
Generative AI tools cannot own patent rights.
Autonomous software cannot qualify as a joint inventor.
Even if an AI system produces technically sophisticated outputs or assists in generating potential solutions, patent rights ultimately depend on identifying the human individuals who contributed to the conception of the claimed invention.
What Counts as a Human Inventive Contribution?
One of the central themes of the revised guidance is the continued emphasis on evaluating human contribution to the conception of each claimed invention.
Simply using AI during research does not disqualify an invention from patent protection. Likewise, merely operating an AI system does not automatically make someone an inventor. The controlling inquiry remains whether one or more natural persons contributed to the conception of the claimed invention.
Human inventive contribution may include situations where a researcher:
Identifies the technical problem to be solved
Designs prompts or iterative inputs that, depending on the facts, contribute to forming the inventive concept
Evaluates multiple AI-generated outputs and recognizes a specific inventive solution
Modifies or refines AI-generated material into the claimed invention
Integrates AI-generated information with independent human insight to develop the final claimed invention
By contrast, merely adopting or implementing an AI-generated output without contributing to the conception of the claimed invention may not satisfy the inventorship requirement. The guidance therefore emphasizes substance over tool use, reaffirming that inventorship is determined by human conception rather than the technology used in the inventive process.
AI Is a Tool, Not an Inventor
A recurring theme throughout the revised USPTO guidance is that artificial intelligence does not alter the legal framework for determining inventorship. Instead, inventorship continues to be defined by human conception under existing patent law principles.
In practice, AI is treated as a technology that may assist in the inventive process, but it does not itself qualify as a legal inventor under the Patent Act. The relevant inquiry remains focused on identifying the natural persons who contributed to the conception of the claimed invention. This approach is consistent with how U.S. patent law has historically treated other technologies that assist in innovation, including:
Computer-aided design (CAD) software
Simulation and modeling tools
Statistical and data analysis systems
High-throughput laboratory and screening equipment
Other computational tools used in research and development
While these tools may significantly influence the development of an invention, they do not themselves contribute to inventorship. The USPTO’s revised guidance recognizes that modern AI systems may operate with greater autonomy and generate more sophisticated outputs than traditional software tools. Nevertheless, it confirms that this increased capability does not alter the governing legal analysis. Inventorship remains grounded in human conception of the claimed invention, regardless of the sophistication of the tools used in the inventive process.
What the Revised Guidance Means
The USPTO’s November 2025 revised guidance does not create a new legal framework for AI-assisted inventions. Instead, it reaffirms that traditional inventorship principles continue to apply, regardless of whether artificial intelligence played a role in the inventive process. Rather than relying on AI-specific examples introduced in earlier guidance, the USPTO now emphasizes application of longstanding principles of conception and inventorship under existing patent law.
This clarification also addresses several common points of confusion that have emerged as AI becomes more deeply integrated into research and development. In practice, the revised guidance underscores that inventorship analysis continues to depend on human contribution to conception, rather than the mere use of AI tools. Common issues addressed in the guidance include:
AI involvement does not eliminate patent eligibility.
An invention is not excluded from patent protection simply because AI contributed to its development. Many AI-assisted inventions remain patentable so long as at least one natural person made a sufficient inventive contribution to the claimed subject matter.
AI users are not automatically inventors.
Merely using an AI tool, issuing prompts, or generating outputs does not establish inventorship. The controlling inquiry remains whether the individual contributed to the conception of the claimed invention.
Prompt engineering is fact-dependent.
Designing prompts may, in certain circumstances, contribute to conception, but it is not automatically sufficient. The determination depends on the specific facts, including the role the prompts played in developing the claimed invention and whether they reflect true inventive contribution.
AI underscores the continued importance of inventorship analysis.
Because AI systems may generate large volumes of technically useful information, careful evaluation of human contribution remains essential in determining inventorship under the Patent Act.
The revised guidance is also firmly grounded in existing judicial precedent rather than introducing new doctrine. The USPTO aligns its analysis with established case law, including:
Thaler v. Vidal (confirming that only natural persons may be named as inventors)
Pannu v. Iolab Corp. (setting standards for joint inventorship)
Burroughs Wellcome Co. v. Barr Laboratories, Inc. (defining conception as the “touchstone” of inventorship)
Ethicon, Inc. v. United States Surgical Corp. (requiring each joint inventor to contribute to conception of at least one claim)
Together, these authorities reinforce a consistent principle: inventorship is determined by human conception, not the tools used in the inventive process. The revised guidance therefore reflects continuity in patent law rather than a doctrinal shift.
Practical Considerations for Inventors and Businesses
For companies integrating artificial intelligence into research and development, the revised USPTO guidance carries important operational and legal implications. While the underlying legal standard for inventorship has not changed, applying that standard in AI-assisted workflows can raise additional factual and evidentiary considerations.
Organizations should therefore focus on strengthening internal processes related to documentation and patent readiness. In particular, companies may consider implementing structured inventorship and disclosure practices such as:
Maintain detailed invention disclosure records: Document each contributor’s role in developing the claimed invention, including technical decision-making and problem-solving contributions.
Track human contributions to AI-assisted outputs: Record how researchers evaluated, selected, or modified AI-generated results, rather than relying solely on raw outputs.
Document key design and engineering decisions: Capture instances where human judgment shaped the invention, particularly where AI was used to generate options or recommendations.
Preserve relevant development records, including prompts where appropriate: In certain circumstances, prompt design and iterative interactions with AI systems may serve as factual context in assessing inventive contribution.
Conduct inventorship reviews before filing patent applications: Early review helps ensure that only proper inventors are named and reduces the risk of correction proceedings or disputes later in prosecution or litigation.
Provide regular training on AI and inventorship principles: Engineering and R&D teams should understand how AI tools interact with patent law requirements and inventorship standards.
These practices are particularly important in industries where AI is heavily integrated into innovation workflows, including biotechnology, pharmaceuticals, medical devices, software development, semiconductor design, aerospace systems, autonomous vehicles, and advanced materials science.
Universities and research institutions face similar considerations. Many institutions may need to revisit invention disclosure procedures and internal intellectual property policies to ensure they adequately address AI-assisted research. Employers more broadly may also need to update internal guidelines governing confidentiality, use of generative AI tools, and employee invention reporting obligations.
From a practical standpoint, inventorship questions are increasingly being addressed earlier in the development lifecycle rather than at the time of patent filing. This proactive approach helps reduce uncertainty and supports more defensible patent applications.
Ultimately, the revised guidance reinforces a straightforward principle: while AI may assist in generating innovative ideas or technical solutions, patent rights continue to depend on identifying the human individuals who contributed to the conception of the claimed invention. Organizations that treat inventorship as a structured, documentation-driven process rather than an afterthought will be better positioned to secure and defend patent protection in an increasingly AI-enabled innovation environment.
If you’re interested in learning more about this topic or how the principles discussed in this article may impact your business, don’t hesitate to contact us at info@patentxl.com or at +1(610)871-2024.
