As artificial intelligence (AI) tools become increasingly integral to the invention process, innovators and companies are leveraging these technologies to accelerate development and refine concepts. From generating initial ideas to optimizing designs, AI can “flesh out” inventions in ways that were previously unimaginable.
However, under U.S. patent law, only natural persons—humans—can qualify as inventors. This principle was reaffirmed in the United States Patent and Trademark Office’s (USPTO) revised inventorship guidance for AI-assisted inventions, issued in November 2025 (https://www.uspto.gov/subscription-center/2025/revised-inventorship-guidance-ai-assisted-inventions).
For this reason, Sherman IP advises clients to prioritize maintaining meticulous documentation to demonstrate substantial human involvement, supervision and direction, and what the attributes of that human involvement were, thereby maximizing their inventions’ eligibility for patent protection.
The Evolving Landscape of AI-Assisted Inventions
The USPTO’s latest guidance rescinds its prior February 2024 directives and emphasizes that no separate or modified legal standard applies to AI-assisted inventions. Instead, the longstanding requirement of “conception” remains central. Conception refers to the formation of a definite and permanent idea of the complete and operative invention in the mind of the human inventor.
AI systems are treated as tools, much like calculators or laboratory equipment, and cannot be named as inventors or joint inventors.
This approach aligns with key court decisions, such as Thaler v. Vidal, 43 F.4th 1207 (Fed. Cir. 2022) which confirmed that inventors must be human beings, i.e. natural persons and not machines.
The guidance clarifies that the Pannu factors—for assessing joint inventorship—apply only among multiple human contributors, not to evaluate AI’s role. See generally, Pannu v. Iolab Corp., 155 F.3d 1344 (Fed. Cir. 1998) The Pannu factors are a set of criteria used to determine whether a human’s contribution to an invention is “significant” enough to warrant naming them as an inventor. They include:
- Nature of the contribution – Was the person’s contribution a substantial, non-obvious contribution to the invention?
- Degree of skill required – How difficult was the contribution to make?
- Amount of time spent – How much effort and time did the person invest?
- Resulting benefit – What tangible or intangible benefit did the contribution produce?
- Other factors – Any additional relevant circumstances.
When AI is used, the focus shifts to proving that a human (not the machine) provided the core inventive spark, such as framing the problem, iterating on solutions, or making critical decisions that go beyond mere oversight.
Without clear evidence of human conception, patent applications risk rejection under 35 U.S.C. §§ 101 and 115 for improper inventorship. Moreover, failure to disclose material AI use could trigger duty of candor violations under 37 C.F.R. § 1.56—which requires patent applicants, inventors, and their representatives to disclose to the USPTO all information known to them that is material to the patentability of the invention, including potentially relevant AI contributions—potentially rendering patents unenforceable.
Importantly, if you use AI to assist in developing your concept, you should still be considered the inventor—as long as you substantially drive the conception. For example, directing AI with specific, inventive prompts, evaluating and refining its outputs, or integrating them into a novel solution demonstrates your significant role; as opposed to the AI driving the conception. For example, you would have trouble being considered an inventor if your involvement is minimal, such as merely posing a generic question to AI and accepting its generated idea without adding your own conceptual input.
The Critical Role of ‘Conception’ in AI-Assisted Innovation
Many innovations begin with a high-level concept, such as identifying a problem or goal. While this can spur an inventor’s creativity, it often represents a general research plan rather than full conception.
Under USPTO standards, inventorship requires a human to form a definite and permanent idea of the complete and operative invention, including the “how”, i.e., the specific, workable details to make the invention function.
For complex or emerging technologies, the operative implementation (e.g., specific methods, processes, or integrations) is key. If a human provides only the broad idea of a general concept, and lets AI generate these details of the implementation of the concept, without significant human input, refinement, or decision-making, the human may not qualify as the inventor for the invention that flows from the general concept. Courts and the USPTO emphasize that mere approval of AI outputs isn’t enough; the core inventive spark must come from the human themself.
This underscores why documentation is essential: Innovators should carefully record how their expertise shaped the development of the invention, from tailored prompts to iterative refinements. Consulting Sherman IP early can help evaluate when your contributions meet the conception threshold, and help you in developing that invention to a sufficiently inventive point, thereby avoiding Patent Office rejections or legal challenges.
Common Pitfalls in AI-Assisted Inventions
AI can be a powerful ally, but over-reliance on AI without substantial human contribution can jeopardize patentability. Here are some key pitfalls to avoid:
- Insufficient Human Conception: If AI generates the core idea and a person simply implements or lightly edits the output, inventorship may be brought into question. For example, prompting AI with “invent a new battery” and adopting the response with minimal modification may not qualify as human conception.
- Enablement Dependencies: If the invention’s concept relies so heavily on AI that it couldn’t be fully conceived or enabled by a human without the AI’s unique, non-replicable contributions, this could raise issues of inventorship when drafting and examining the claimed invention in the patenting process. While enablement (under 35 U.S.C. § 112) requires a specification to teach skilled artisans how to make and use the invention, excessive AI dependence might imply the human lacked true conception, inviting scrutiny during prosecution or litigation.
- Failure to Disclose AI Involvement: Failing to document or disclose material AI involvement (e.g., if AI introduced key elements you didn’t conceive) violates the duty of candor, potentially leading to issues of unenforceability.
- Data Privacy and Confidentiality Risks: Using public AI tools without safeguards could inadvertently disclose proprietary ideas, complicating patent priority or enabling competitors. Remember publication of a claimed invention in advance of filing a patent application jeopardizes that protection under the patent laws.
Understanding these risks helps innovators ensure AI strengthens—rather than undermines—their patent strategy and their potential for obtaining and enforcing strong, valid patents.
Why Documentation Matters
In an era where AI can generate complex outputs from simple prompts, distinguishing human ingenuity from automated assistance is paramount. Clear documentation serves multiple purposes:
- Establishing Inventorship: Records can demonstrate that humans conceived the invention, not just approved AI-generated results.
- Supporting Patent Prosecution: While the USPTO presumes the named inventor(s) is/are correct, Patent Examiners may inquire about AI involvement if inventorship appears unclear or if the patent application suggests heavy reliance on AI without evident human conception. In such cases, a rejection could occur under §101 (ineligible subject matter due to lack of a human inventor) or §115 (defective oath or declaration). Solid documentation can preempt or resolve such issues, making rejections less likely when human contributions are well-evidenced from the outset.
- Defending Against Validity Challenges: More commonly, the need for this documentary evidence arises in infringement litigation, where a defendant might challenge the patent’s validity by arguing that the named inventor did not substantially and sufficiently participate in the inventive process—claiming, for instance, that AI was the true “conceiver.” In post-grant proceedings or court, robust, contemporaneous records of human involvement can effectively counter such invalidity arguments, preserving the patent’s enforceability.
- Complying with Duties: Applicants, inventors, and patent practitioners have obligations under 37 C.F.R. § 1.56 to disclose information material to patentability, including AI’s role if it affects inventorship claims.
Best Practices for Documenting AI Use
To safeguard your intellectual property, Sherman IP recommends implementing structured documentation practices. To the extent you systemically rely on AI across a portfolio of patent applications, our attorneys can review, revise and advise you on systems to best document your human conception contributions (and advise you where support can be added for best practice.) Some practices we recommend are:
- Update Invention Disclosure Processes: Revise internal Invention Disclosure forms to explicitly ask inventors about AI tools used, including the type (e.g., generative AI like ChatGPT, Grok or specialized models), prompts provided, and outputs generated. Require inventors to detail their pre-AI conceptualization, such as problem identification or hypothesis formulation, as well as their post or continuing conceptualization that may occur while interacting with AI during the development process.
- Log Iterative Interactions: Maintain timestamped records of all AI engagements. This includes:
- Initial and refined prompts showing human direction.
- AI responses and how humans evaluated, modified, or rejected them.
- Decisions on integration, such as selecting specific features or overcoming limitations identified by AI.
- Track Human Contributions Per Claim: For each patent claim, log and document the human role in its development. For instance, if AI suggested alternatives, note how a human chose or combined them to form the operative invention. This ensures every claim reflects significant human input. Usually, the patent claims are drafted at the patent application process which follows conceptualization. A detailed log will support the claim development here.
- Preserve Version History and Supporting Materials: Use tools like Git or secure databases to version-track documents, sketches, prototypes, and emails. Include notes on human thought processes, meetings, and refinements to illustrate conception.
- Train Teams and Implement Policies: Educate research and development teams on inventorship standards and the importance of real-time documentation. Establish company policies prohibiting naming AI as an inventor and requiring review of AI-assisted drafts for accuracy and confidentiality. As part of patent harvesting sessions we implement with our systemic invention clients, we engage in this educational process and advise on company policies that may be applied or revised. Also, vet AI platforms for data privacy to avoid inadvertent disclosures under 37 C.F.R. § 11.106(d).
- Disclose Material AI Contributions When Necessary: If AI introduced elements that could affect inventorship or patentability, consider submitting an Information Disclosure Statement (IDS) explaining the AI’s role and amending the claims where necessary. Proper disclosure can satisfy the duty of candor without undermining valid inventorship claims.
Practical Advice for Our Clients
As AI adoption grows, proactive documentation and recordkeeping will be key to unlocking patent protections while mitigating risks. At Sherman IP, we work with clients to integrate these practices into their innovation workflows– from initial concept through patent filing. By emphasizing human-centric records, you not only comply with USPTO requirements but also strengthen your competitive edge.
If your team is using AI in development, Sherman IP can help evaluate your processes and ensure your inventions remain protected as human achievements in an AI-driven world.
Sherman IP LLP
Protecting Your Vision, Securing Your Future