New guidance on AI-assisted inventions from the U.S. Patent and Trademark Office (USPTO) eases patentability restrictions, allowing humans to use AI systems such as large language models (LLMs) to develop new ideas while still being able to patent those inventions.
This guidance comes as the USPTO undergoes a substantial pro-patent shift under the Trump administration. For example, the USPTO has substantially reduced the availability of IPRs to patent defendants and has established new PTAB precedent that signals a potential easing of subject-matter eligibility rejections for AI/ML inventions.
In its new guidance, the USPTO effectively says that humans may use AI systems without potentially losing human inventorship. The guidance analogizes AI systems to traditional equipment used to develop ideas, like laboratory equipment, software tools, or research databases, and argues that a human who uses an AI system to develop an idea is similarly still the one who “conceived” the idea. Thus, the USPTO’s new policy lessens the likelihood that a patent application will be rejected for lacking a human inventor in cases where the human used an AI system when developing the claimed invention.
The USPTO’s previous guidance on AI-assisted inventions required “significant human contributions” to an invention. This approach, tied to the three factors for inventorship set forth in Pannu v. Iolab Corp., 155 F.3d 1344 (Fed. Cir. 1998), raised the possibility that a human could over-rely on an AI system during development such that they no longer constitute an inventor. The USPTO’s new guidance rescinds this policy in favor of their new pro-patent policy.
While the USPTO’s new guidance is binding on patent examiners reviewing patent applications, it is not binding on courts reviewing asserted patents. No court has considered the question of how much human contribution, if any, is required for a human to constitute an inventor when using an AI system, so there is some uncertainty about where the courts will land on this issue. It is also worth noting that the USPTO’s new relaxed policy on the use of AI systems is inconsistent with the Copyright Office’s approach to hybrid AI-human collaborative works. This discrepancy demonstrates the differing views that IP attorneys have on how AI systems can impact IP ownership.
Those concerned with ensuring their ideas are patent-protectable while using AI systems should confirm that a human participated in the “conception” of the invention. Acceptable uses of AI may include a human iteratively prompting an AI system to refine the output, meaningfully revising the output of the AI system when implementing the idea, or specially training the AI model used to develop the invention. Click here for more tips on maximizing your invention’s patentability.