Artificial intelligence (AI) has quickly moved from the lab into everyday life, powering tools that help us search, write, analyze data, code, illustrate and make faster decisions. It streamlines routine tasks, surfaces insights hidden in complex information and enables new products and services across industries. On the other hand, indiscriminate AI use can expose organizations to potential intellectual property infringement and other issues. Understanding what AI can—and cannot—do today is essential to capturing its benefits while managing its risks.
As background, AI is typically “trained” by feeding massive datasets (e.g., text, images, video, computer code, etc.) into Large Language Model (LLM) algorithms that analyze that data for patterns, adjusting internal parameters (or “weights”) to improve accuracy over time. Where do the massive datasets originate? Typically, it is crawled or scraped from “the Internet” meaning, uncountable websites, digitized books, code repositories, and more. One immediate question is: Does the AI developer have the right (i.e., copyright license or the like) to do that? The answer, as historically illustrated in Bartz v. Anthropic PBC, 3:24-cv-05417, (N.D. Cal.), is likely no, and Anthropic paid a $1.5 billion settlement for downloading pirated works.
But what about patentable inventions? Should AI be used to assist there? As with any evolving technology, the answer is not definitive, but with appropriate forethought and care AI may be used as a helpful tool under the following non-exhaustive guidelines.
First, AI cannot be a named inventor on a patent application. Only “natural persons” can be inventors. Thaler v. Vidal, 43 F.4th 1207 (Fed. Cir. 2022). However, under current U.S. Patent & Trademark Office guidelines (available at 90 FR 54636, issued Nov. 28, 2025) AI systems can be used as instruments by human inventors and are analogous to laboratory equipment, computer software, research databases or any other tool that assists the inventive process.
Second, to avoid “abstract idea” patent ineligibility issues, the invention must “do more” than claim the application of AI to a new data environment. In other words, if the claimed invention merely uses “generic AI” to perform a process better, faster or more efficiently than a human could, it is merely an unpatentable abstract idea. Recentive Analytics, Inc. v. Fox Corp., 134 F.4th 1205 (Fed. Cir. 2025).
Third, there are significant disclosure issues that should be evaluated when using AI to assist with inventing. One should not assume that chats with AI are private. The AI tool provider may (likely will) use your chats and inputs to further train its AI and could disclose your nascent invention to another (potentially a competitor) entering a similar prompt. Likewise, the “answer” an AI tool provides to your invention queries and prompts may originate from data crawled from the Internet and rightfully belonging to someone else which may impact your ability to patent your invention and could present infringement problems.
Parsons Behle & Latimer attorneys can assist you in assessing your AI best practices with respect to all your intellectual property needs.
