USPTO Issues Superseding Guidance On Inventorship For AI-Assisted Inventions

Key Takeaways:

• Practical impact. Lower, more flexible threshold likely makes it easier to establish human inventorship on AI-assisted applications.
• Still case-by-case, but less structured. The USPTO removes the prior multi-factor framework, enabling broader arguments (e.g., interpreting ambiguous AI results and choosing a development path).
• Expanded pathways to inventorship. Selection/interpretation of AI results, recognizing and pursuing valuable AI outputs, customizing AI for the task, and crafting insightful prompts may now suffice.

In early 2024, the U.S. Patent and Trademark Office (USPTO) released detailed guidance for determining inventorship of AI-assisted inventions. That guidance emphasized that only natural persons could be inventors and provided a set of five "guiding principles" to help determine inventorship in the context of AI. Those principles offered a granular, case-by-case framework for evaluating the roles of individuals who interact with or develop AI systems in the course of inventive activity. However, on November 28, 2025, the USPTO issued new guidance explicitly rescinding and replacing the 2024 Guidance, signaling a significant shift in the agency's approach.

Summary of the 2025 USPTO Guidance

The 2025 Guidance takes a notably streamlined approach. While it reaffirms the core principle that only natural persons may be named as inventors, it dispenses with the detailed, multi-factor analysis of the prior guidance. Instead, the new guidance focuses on a simplified standard: a human is an inventor if they make a "significant contribution" to the conception of the claimed invention, consistent with longstanding case law. The USPTO has withdrawn the five guiding principles and the extensive hypotheticals that characterized the 2024 Guidance. Rather, the new guidance emphasizes flexibility and a lower threshold for establishing inventorship, making clear that human involvement in directing, selecting, or interpreting AI outputs may be sufficient to support inventorship, provided there is a meaningful contribution to the inventive concept.

Comparison: 2024 vs. 2025 Guidance

The shift from the 2024 to the 2025 Guidance is substantial:

• Idea Generation Now Need Not Be Attributable to a Natural Person
  Under the 2024 Guidance, only a natural person who makes a substantial contribution to the conception of the invention can be named as an inventor. Merely using an AI tool, without more, was not enough. Under the new guidance, a researcher who uses an AI system to generate a novel compound and then selects one promising candidate for further study could now be considered an inventor, even if their primary contribution was the selection and interpretation of the AI's output—something that may not have sufficed under the 2024 standard.
• Merely Recognizing and Appreciating AI Output May Now Be Sufficient
  Under the 2024 Guidance, simply recognizing the value or utility of an AI-generated result did not qualify as inventorship; a human had to make a more direct intellectual contribution to conception. Under the newly-announced guidance, if a scientist reviews a set of AI-generated solutions and identifies one as having unexpected utility, that act of recognition and subsequent pursuit may be enough to establish inventorship, even if the scientist did not originate any portion of the solutions.
• Routine Programming or Maintenance of AI May Now Be Inventive
  Under the old guidance, individuals who merely programmed, maintained, or trained an AI system, without contributing to the inventive concept, were not inventors. Under the new approach, a developer who customizes an AI model for a specific inventive task and then interprets its outputs in a way that leads to a patentable invention could now be considered an inventor, even if their primary role were technical rather than conceptual.
• A Prompt Or Input From A Human May Now Suffice
  The 2024 Guidance made clear that providing a prompt or input to an AI system, without more, was not enough to establish inventorship; the human had to contribute to the inventive concept itself. With the 2025 Guidance, a user who crafts a particularly insightful prompt that leads the AI to generate a novel and non-obvious solution may now be able to claim inventorship, as the act of directing the AI is given more weight.
• Inventorship Is Still a Fact-Specific Inquiry
  The fact-specific nature of inventorship was emphasized in the 2024 Guidance, but the guidance provided a structured, multi-factor analysis to aid in the determination. The new guidance removes much of this structure, allowing for broader arguments. For instance, a team member who interprets ambiguous AI results and decides which direction to pursue in further development may now be able to claim inventorship, even if their contribution would have been considered too attenuated under the 2024 framework.

Conclusion

The USPTO's 2025 Guidance marks a clear departure from the detailed, principle-driven approach of 2024. By rescinding the prior guidance and adopting a more flexible, less prescriptive standard, the USPTO has arguably made it easier to establish human inventorship for AI-assisted inventions. Applicants and practitioners should be aware of this shift and consider how the new, streamlined approach may impact the evaluation and assertion of inventorship in future patent filings involving AI.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.