How Does AI-Assisted Patent Inventorship Guidance Define Human Contribution?

USPTO Clarifies Human Inventorship Standards

The USPTO’s revised guidance treats artificial intelligence as a powerful tool rather than an inventor. A patent application must name at least one natural person who made a significant contribution to the claimed invention. Merely prompting an AI system, reviewing its output, or recognizing its usefulness generally does not establish inventorship. Instead, the human must shape the problem, direct the system, evaluate alternatives, and contribute to the conception of the final claimed solution in a meaningful way.

Also worth reading: Can AI-Assisted Inventorship Errors Still Be Corrected Under Section 256? · What Should an AI-Assisted Inventorship Documentation Checklist Include? · What Are the USPTO AI Inventorship Documentation Requirements for Patent Filings?

This standard focuses on the human’s intellectual contribution, not on who performed mechanical tasks. When AI helps generate compounds, designs, or code, the inventor is the person who conceived the inventive concept and controlled its development. Clear records of human decisions, experiments, and refinements can support proper inventorship. The guidance also reminds applicants that incorrect inventorship may be corrected under Section 256 when possible, reinforcing that AI-assisted innovation remains patentable when a human’s creative role is documented and central.

AI Tools Versus Named Inventors

The USPTO's revised guidance clarifies that AI systems cannot be named as inventors, but inventions created with AI assistance may still be patentable when a human makes a significant contribution to the conception. The guidance emphasizes that the human must contribute meaningfully to the claimed invention, not merely operate the tool or recognize its output. This aligns with the Thaler v. Vidal ruling affirming that only natural persons can be inventors under U.S. patent law.

Practitioners must document the human's intellectual role throughout the inventive process, especially in fields like AI drug discovery where models generate candidate compounds. The guidance also addresses correction of inventorship under Section 256, allowing errors to be fixed without invalidating patents when the human contribution is properly established. Ultimately, the framework treats AI as an instrument, ensuring patents reward genuine human ingenuity while accommodating modern research tools.

Correcting Inventorship Under Section 256

The guidance treats artificial intelligence as a tool rather than an inventor, so patentability depends on whether a natural person made a significant contribution to the conception of the claimed invention. A human who frames the problem, designs experiments, selects or refines AI outputs, or integrates results into a complete and operative invention may qualify, while merely prompting a system or recognizing its output generally does not. This distinction keeps inventorship tied to human creativity even when machine learning accelerates research.

Section 256 provides a mechanism to correct inventorship when the record does not accurately reflect those human contributions, including errors involving AI-assisted work. Practitioners should document who conceived what, review claims against actual human input, and seek correction before enforcement or licensing. Clear records help preserve validity and reduce disputes over whether the named inventors truly shaped the invention.

Drug Discovery And Trade Secrets

The USPTO’s revised guidance treats AI as a tool, not an inventor, and requires a natural person to make a significant contribution to conception of each claim. Human contribution may include designing the research problem, selecting or training models, interpreting outputs, and refining claims, but merely prompting or accepting AI-generated results is generally insufficient. In drug discovery, this means scientists must shape hypotheses, validate targets, and direct experimental follow-up rather than rely on automated suggestions alone.

The guidance also emphasizes that inventorship is claim-specific, so each claim needs at least one human who significantly contributed to its conception. Where AI accelerates compound screening or structure prediction, the human role in framing questions and judging results remains central. Companies should document decisions, data choices, and validation steps to support inventorship while protecting trade secrets through controlled access, confidentiality agreements, and careful disclosure strategy.

Thaler Vidal And Patent Eligibility

The USPTO’s guidance on AI-assisted inventions centers on whether a natural person made a significant contribution to conception, not whether an AI system generated output. A human must do more than recognize a problem or own the machine; they must shape the invention through judgment, selection, or refinement. The guidance treats AI as a tool, like laboratory equipment, while requiring human intellectual contribution to the claimed invention. It also emphasizes that inventorship depends on the specific claims, so different claims may reflect different human contributions.

Under this approach, routine prompting, accepting AI output, or merely supervising the process is generally insufficient. A human contributor should meaningfully direct the invention’s development, such as designing experiments, choosing parameters, or integrating results into a patentable concept. The guidance also reminds practitioners that naming only humans remains required, and that incorrect inventorship can be corrected under Section 256 when errors are not deceptive. Ultimately, the standard preserves patent law’s human-centered foundation while accommodating AI as an assistive technology.

Human Inventor Versus AI Tool

USPTO guidance treats AI as an assistive tool, not an inventor. Patentability depends on a natural person’s significant contribution to conception for each claimed invention, whether through designing the system, crafting prompts, or refining outputs. Mere ownership, oversight, or recognition of AI-generated results does not establish inventorship, so teams should document human creative steps carefully.