## The Core Rule: Natural Persons Only The United States Patent and Trademark Office has made clear that inventorship under 35 U.S.C. § 100 remains tethered to natural human beings. An AI system, regardless of how much it contributed to the conception of a claimed invention, cannot be named as an inventor. This position was reinforced in the wake of the Thaler v. Vidal decision, where the Federal Circuit held that an inventor must be a natural person, and the Supreme Court declined to hear the case in 2025. For practitioners, this means every inventor listed on a patent application must be a human being who made a contribution to the conception of at least one claim. The USPTO's 2024 guidance on AI-assisted inventions did not alter this fundamental rule but instead provided a framework for identifying when a human inventor's mental contribution is sufficient. Practitioners should treat the AI tool as a drafting or analysis aid, not as an inventor, and structure the application to reflect the human inventor's specific intellectual contributions to the problem-solving process.
## How the USPTO Defines Inventorship for AI-Assisted Inventions The USPTO's guidance on AI-assisted inventions, published in early 2024, sets out a two-part inquiry for determining whether a human inventor exists. First, the practitioner must identify the specific problem the human inventor sought to solve. Second, the practitioner must determine whether the human inventor recognized the AI output as a solution to that problem and made a meaningful contribution to the conception of the final claimed invention. If the human merely presented a problem to the AI and accepted the first output without further intellectual input, that person may not qualify as an inventor. Conversely, if the human inventor evaluated, modified, or combined the AI output with other knowledge to arrive at the claimed invention, inventorship is properly attributed. The guidance emphasizes that the use of AI does not automatically strip a human of inventorship, but it also does not permit naming the AI itself. Practitioners should document the human inventor's specific cognitive steps in the invention process, including any post-AI-output analysis, selection, or refinement that led to the final claims.
Also worth reading: What are the current rules for AI patent inventorship and how do they apply to agentic AI systems? · What are the current requirements for patent eligibility for machine learning models in the United States? · What are AI patent co ownership best practices for joint inventorship and licensing?
## Practical Steps for Drafting AI-Assisted Patent Applications Practitioners should begin by conducting a detailed interview with the human inventor to map the invention process from problem identification through final claim drafting. This interview should capture the specific role the AI tool played, whether the tool generated raw output that was subsequently evaluated and modified, and the nature of the human inventor's intellectual contributions at each stage. The patent application should include a detailed description of the human inventor's problem-solving approach, with the AI tool described as a resource or instrument rather than a co-creator. Claims should be drafted to reflect the human inventor's specific conceptual contributions, and the specification should explain how the human inventor recognized the AI output as a solution and what additional steps were taken to arrive at the final invention. Practitioners should also prepare an inventor oath or declaration that accurately names only the human inventors and avoids any reference to AI systems as inventors. This documentation strategy reduces the risk of a rejection based on improper inventorship and provides a clear evidentiary record if the application is later challenged.
## Comparison Table: Inventorship Standards Across Jurisdictions
| Feature | United States | Europe (EPO) | China (CNIPA) |
|---|---|---|---|
| Inventor must be a natural person | Yes, per 35 U.S.C. § 100 | Yes, per EPC Article 60 | Yes, per Patent Law Art. 15 |
| AI system can be named as inventor | No | No | No |
| Guidance on AI-assisted inventions | 2024 USPTO guidance issued | EPO updated examination guidelines referencing AI | CNIPA has not issued specific AI inventorship guidance |
| Human contribution threshold | Must contribute to conception of at least one claim | Must make an inventive contribution to the claimed invention | Must be the actual creator of the invention |
| Oath/declaration requirement | Required, naming only human inventors | Required, naming only human inventors | Required, naming only human inventors |
## When to Act and What to Do Now Practitioners should take immediate steps to audit any pending or pending-pending patent applications that involve AI-assisted inventions. This audit should verify that every named inventor is a natural person who made a qualifying contribution to the conception of the claims. For new applications, practitioners should implement a standardized intake process that captures the human inventor's role in the invention process before any AI tools are used in drafting. If an application has already been filed with an AI system listed as an inventor or with a human inventor who did not make a qualifying contribution, the practitioner should consider filing a correction under 37 CFR § 1.48 or preparing a new application with proper inventorship. The cost of correcting inventorship errors after filing can be substantial, including the risk of invalidation in litigation or post-grant review proceedings. Practitioners should also monitor the USPTO's ongoing AI agenda, which includes examination of the Office's own AI tools and guidance for practitioners, as additional rules or examples may be issued in 2026 that further clarify the inventorship framework.
## Cost Considerations and Pricing Implications The cost of ensuring proper inventorship for AI-assisted inventions is primarily a function of the practitioner's time spent on inventor interviews, documentation, and application drafting. A typical inventor interview for an AI-assisted invention may take 60 to 90 minutes and cost between $300 and $800 depending on the practitioner's hourly rate and the complexity of the technology. Drafting a specification that clearly delineates the human inventor's contributions from the AI tool's role may add 2 to 5 hours of drafting time compared to a conventional application, translating to an additional $500 to $2,000 in legal fees. If inventorship corrections are needed after filing, the cost of a supplemental examination or a new application can range from $5,000 to $15,000 or more, depending on the scope of the claims and the number of jurisdictions involved. These costs are modest compared to the potential consequences of an inventorship challenge, which can result in the invalidation of an entire patent portfolio. Practitioners should factor these costs into their engagement letters and project budgets for AI-related patent work.
## The Evolving Regulatory Picture and What Practitioners Should Watch The USPTO's AI agenda continues to evolve, with the Office examining its own AI tools and guidance for practitioners. The 2024 inventorship guidance for AI-assisted inventions is not the final word, and practitioners should expect further updates as the technology and the case law develop. The Supreme Court's refusal to hear the Thaler case in 2025 effectively closed the door on AI inventorship in the United States for the foreseeable future, but legislative proposals in Congress could alter the statutory framework. In Europe, the EPO's approach remains consistent with the requirement that inventors be natural persons, and the China National Intellectual Property Administration has not yet issued specific guidance on AI inventorship, though the general principle that only natural persons can be inventors is well established. Practitioners working on cross-border AI patent portfolios should coordinate inventorship determinations across jurisdictions to ensure consistency and avoid the risk of invalidation in any single country. The International Patent Classification system is also beginning to reflect AI-related technologies, and practitioners should stay informed about these developments as they plan their patent strategies for 2026 and beyond.