The Core Question: Can an AI Be Named as an Inventor?
Under current United States patent law, only natural persons can be named as inventors on a patent application. This principle was affirmed by the U.S. Court of Appeals for the Federal Circuit in Thaler v. Vidal (2022), which held that the Patent Act's use of the word "individual" restricts inventorship to human beings. The USPTO has consistently maintained this position, rejecting applications where an artificial intelligence system was listed as the sole inventor. For agentic AI systems that autonomously generate designs, algorithms, or technical solutions, this creates a fundamental tension. The entity or company deploying the agentic AI may believe the system itself deserves recognition, but the law does not currently permit it. Practitioners must therefore trace the human contributions that led to the AI's output to identify proper inventors. This requires understanding both the technical architecture of the agentic system and the legal standards for conception. The question is not whether the AI contributed to the invention, but whether a human exercised sufficient creative control and intellectual dominion over the result. Until Congress amends the Patent Act or courts revisit the issue, the human-centric inventorship framework remains the governing standard.
Also worth reading: What is the current USPTO guidance on AI patent inventorship as of 2024, and how should inventors and practitioners comply with these requirements? · How can enterprises approach agentic AI legal risk mitigation while developing autonomous systems? · What are privacy-preserving observability frameworks and how do they secure agentic AI systems?
How Agentic AI Differs from Traditional AI Tools in Patent Law
Agentic AI systems operate with a degree of autonomy that distinguishes them from conventional AI-assisted design tools. These systems can set sub-goals, iterate on solutions, and execute multi-step workflows without continuous human direction. In traditional AI-assisted invention, a human provides a prompt, reviews outputs, and makes iterative refinements, establishing a clear chain of human conception. With agentic AI, the system may generate a final solution that no single human could have anticipated or manually assembled. This autonomy complicates the analysis of who conceived the claimed invention. The USPTO has not yet issued specific guidance addressing agentic AI inventorship, though its broader AI agenda includes examining how AI tools affect patent prosecution. The distinction matters because the level of human involvement determines whether a named inventor can satisfy the legal requirement of conception. Conception, under patent law, is the formation in the mind of the inventor of a definite and permanent idea of the complete and operative invention. If a human merely initiates an agentic process and accepts its output, courts and examiners may question whether true conception occurred. The gap between human intent and machine execution is where most inventorship disputes will arise.
Practical Steps for Identifying Inventors in Agentic AI Workflows
Patent practitioners should begin by documenting the entire development chain when an agentic AI system contributes to an invention. This includes recording the initial problem statement, the parameters set by human operators, the training data used, and the specific human interventions that shaped the final output. Each person who contributed to the conception of at least one claim element should be evaluated for inventorship. The key test is whether the individual made a meaningful intellectual contribution to the claimed solution, not merely whether they pressed a button to start the AI. Companies deploying agentic AI should establish internal protocols that flag potentially patentable outputs and trigger a formal inventorship review before any application is filed. Legal teams must coordinate with engineers to reconstruct the human decision points that guided the agentic system. Failure to identify all rightful inventors can lead to patent invalidity challenges under 35 U.S.C. § 256, which permits correction of inventorship but only with the consent of the parties involved. A res judicata bar may prevent correction efforts if a prior proceeding has already determined inventorship. Practitioners should also monitor the evolving guidance from the USPTO, which has signaled interest in AI-related patent issues through its AI agenda and practitioner tools.
Comparison Table: Human-Only vs. AI-Assisted Inventorship Scenarios
| Feature | Human-Only Inventorship | Agentic AI-Assisted Inventorship |
|---|---|---|
| Legal standard | Clear: natural person conceived the invention | Unclear: human contribution to AI output must be traced |
| USPTO acceptance | Routinely accepted | Subject to heightened scrutiny and potential rejection |
| Documentation burden | Standard patent drafting | Requires detailed logs of AI inputs, parameters, and human interventions |
| Risk of invalidity | Low if named inventors are correct | High if human conception is insufficiently documented |
| Correction under §256 | Straightforward with consent | Complicated by res judicata and lack of precedent |
| Global variation | Generally consistent among major offices | Varies widely; some offices more permissive than others |
One of the most frequent errors is naming the AI system or its developer as an inventor, which directly violates the statutory requirement for natural person inventorship. Another common mistake is failing to document the human creative contributions that shaped the agentic AI output, leaving the company vulnerable to inventorship challenges during litigation or post-grant proceedings. Some organizations assume that because they own the AI system, they automatically own any inventions it generates, but ownership and inventorship are legally distinct concepts. A company may own the AI tool without being the inventor of what it produces. Practitioners also err by relying on boilerplate inventorship clauses in employment or contractor agreements that do not address AI-generated works. These agreements may not cover situations where an agentic AI system produces an unexpected result that no human explicitly directed. The failure to update inventorship policies to address autonomous AI systems creates significant exposure. Companies should also avoid filing applications with overly broad claims that implicitly attribute conception to the AI system rather than to specific human contributors.
When to Act: Timing and Strategic Considerations
Organizations should initiate inventorship reviews as soon as an agentic AI system produces a potentially patentable output, ideally before any public disclosure or publication. The on-sale bar under 35 U.S.C. § 102(a)(1) means that public use or sale of an AI-generated invention can forfeit patent rights if a timely application is not filed. Given the speed at which agentic AI systems can iterate and produce results, companies may face a compressed timeline between invention and disclosure. Patent applications should be filed within the statutory grace period where available, but relying on grace provisions is risky and should not substitute for prompt filing. Internal review processes should include a decision point at which legal counsel evaluates whether the human contributions meet the conception threshold. If the documentation is insufficient to establish human inventorship, the company must decide whether to invest additional resources in tracing the inventive chain or to forgo patent protection for that output. The cost of a delayed or missed filing opportunity can far exceed the cost of maintaining robust AI inventorship protocols from the outset.
Cost and Pricing Considerations for AI-Related Patent Prosecution
The cost of prosecuting a patent application involving agentic AI inventions is generally higher than for conventional inventions due to the additional documentation and analysis required. A typical patent application in the United States costs between $10,000 and $30,000 in legal fees, but AI-related applications may add 20 to 40 percent to that range because of the need for technical expert consultations and detailed inventorship analyses. USPTO filing fees for a utility application currently start at approximately $400 for a small entity, with additional fees for claims, search, and examination. Post-grant proceedings such as inter partes review can cost $300,000 or more per proceeding, making correct inventorship from the outset a critical cost-avoidance measure. Companies should budget for ongoing inventorship audits as their use of agentic AI expands, since each new application may present unique factual questions about human contribution. The cost of correcting inventorship after issuance through a §256 proceeding is unpredictable and may require litigation, which can easily exceed $500,000 in legal fees. Investing in proper protocols upfront is significantly less expensive than defending inventorship challenges later.
Global Perspectives and the Outlook for Agentic AI Inventorship
Patent offices around the world are grappling with how to handle AI-generated inventions, and their approaches vary considerably. The United States and the European Patent Office currently require a human inventor, while some other jurisdictions have not yet addressed the question directly. The UK Intellectual Property Office and the Australian Patent Office have considered the issue in specific cases, with outcomes that have not established a clear global consensus. The World Intellectual Property Organization has initiated discussions on AI and intellectual property policy, but no binding international framework exists for agentic AI inventorship. Companies with global patent portfolios must navigate these divergent approaches and may need to tailor their inventorship practices by jurisdiction. The USPTO's AI agenda suggests that the office is actively studying the impact of AI on patent law, and rulemaking or guidance updates may emerge in the coming years. Organizations should monitor developments from the USPTO, the European Patent Office, and WIPO to anticipate changes in inventorship requirements. The trajectory of judicial decisions, particularly any further appeals in the Thaler case or new cases involving agentic AI, will shape the legal landscape for years to come.