Patenting for Inventors Ep. 154 - Can AI be an Inventor? Guidance from the Patent Office

Illustration by @max_gps

Can AI be an Inventor? Guidance from the Patent Office

Podcast Transcript:

Hello and welcome to the Patenting for Inventors podcast, where I make patents less of a puzzle and more of a process. My name is Adam Diament, a registered patent attorney and partner at the law firm of Nolan Heimann in Los Angeles, California. This episode is “Can AI Be an Inventor? Guidance from the Patent Office.”

 Can AI Be an Inventor?

Today, we’re diving into a question that’s not only timely but also a little weird: can artificial intelligence be considered an inventor?

 Spoiler alert: the answer is still no. AI may be doing some pretty impressive things, but the U.S. Patent and Trademark Office—just like the courts—has made it clear that it won’t be handing out inventorship credit to AI systems anytime soon. But that’s only part of the story. While AI can’t be listed as an inventor, it can assist in the inventive process. And that’s where things get legally and practically complicated.

AI’s Role in Innovation

AI is everywhere. It’s generating music, drafting documents, coding apps, and even helping scientists discover new chemical compounds and treatments. Tools like large language models and neural networks are becoming key collaborators in R&D. So, naturally, the question comes up: if AI helps you come up with something novel, is the invention still patentable? And if so, who gets credit as the inventor?

USPTO Guidance on Inventorship

To help address this, the USPTO released formal guidance in February 2024, later reinforced in a broader September 2024 presentation. The message is consistent: only natural persons can be inventors under U.S. law. This has always been the case, but the new guidance is the USPTO’s way of saying, “Look, AI isn’t going away, so let’s talk about how this works when machines are part of the process.”

The “Significant Contribution” Test

Under the updated guidance, an invention developed with AI is still potentially patentable—as long as a human being made a significant contribution to the invention. That phrase, “significant contribution,” comes directly from the Federal Circuit’s decision in Pannu v. Iolab, and it’s now the cornerstone of how inventorship is evaluated for AI-assisted inventions.

Practical Application of the Guidance

So what does that mean in practical terms? Suppose you’re a biochemist using an AI tool to suggest new protein sequences. You feed the AI some parameters, it suggests five options, and you pick one to test based on your experience and knowledge. If your decision-making added intellectual value—like understanding which option has therapeutic potential—then you likely made a significant contribution. You’re the inventor. But if the AI not only suggested the sequence but also modeled its function and told you exactly what to do—and you just followed its output without really contributing to the inventive step—then you probably didn’t contribute enough under the law. That could mean your patent application lacks a valid inventor and is therefore not enforceable.

Clarifying the Role of AI

Where this guidance is helpful is in clarifying that AI is, legally speaking, a tool—not a creator. This prevents companies from listing AI systems as inventors, which would render a patent invalid. It also helps patent practitioners and inventors focus on what does count: human insight, judgment, and conception. But just because the law draws a clear line doesn’t mean everything is black and white in practice.

Gray Areas and Edge Cases

Consider a software engineer who uses generative AI to design a new user interface. If they accept the AI’s output without modification or understanding, it’s hard to argue they made a significant contribution. But what if they tweak one layout detail? Or interpret the AI’s proposal in a way that solves a long-standing usability issue? Are they then an inventor? According to the guidance, it’s all about the nature of the contribution, not just how much work was done. These are the kinds of edge cases that will require careful analysis—often on a claim-by-claim basis.

Collaborative Inventorship and Claim Requirements

The USPTO also acknowledges situations where multiple people are involved in the inventive process. One team member might configure and prompt the AI, another might analyze the results, and a third might craft the claims. Who’s the inventor? The answer depends on who made a significant contribution to the conception of the claimed invention. And importantly, the guidance says not all inventors need to contribute to every claim—but every claim must include at least one human who did.

Disclosure Considerations

Another important point is disclosure. While there’s currently no requirement to proactively disclose that AI tools were used unless the USPTO asks under a formal rule. If an examiner suspects that AI did the inventing and no valid human inventor is named, they can challenge the application. That could lead to a rejection—or worse, invalidation down the road. So while you’re not required to say “Hey, we used ChatGPT to brainstorm this,” you are required to know whether there was meaningful human inventorship and be able to justify it.

USPTO Examples

To illustrate the issues, the USPTO has published several examples, including one involving the development of a therapeutic compound and another involving a mechanical component for a remote-control car. These examples show how a human’s analytical or creative decisions—even when working alongside AI—can make or break the inventorship analysis. You can find those examples on the USPTO’s AI resource page.

Applicability Across Patent Types

It’s also worth mentioning that this guidance applies to all types of patents—utility, plant, and design—and to all applications, including those filed before the guidance was issued. So if you’re sitting on a pending application involving AI, now might be a good time to double-check those inventor declarations and make sure everything lines up with the new expectations.

Connection to Patent Eligibility

One thing this new guidance doesn’t do is change the rules about what subject matter is patentable. But it does intersect with those rules in important ways, especially when the invention involves AI methods themselves. A separate subject matter eligibility update in 2024 provides examples of how AI-related claims should be analyzed under Step 2A of the patent eligibility test, focusing on whether the claim improves technology or merely implements abstract ideas. But that’s a topic for another episode, some of which I’ve covered before.

Final Thoughts

To sum it all up: AI can assist, but it can’t invent. If you’re using AI in your innovation process, be prepared to explain what you, the human, contributed. That contribution must be meaningful and directed at the conception of the invention. If you didn’t add enough, you might not qualify as an inventor—and that could make your patent vulnerable.

Outro

Thanks for listening to The Patenting for Inventors Podcast. If you need help navigating inventorship, drafting claims, or any other questions about intellectual property such as patents, trademarks, copyrights, or business matters, give me a call at 424-281-0162.

Until next time, I’m Adam Diament, and keep on inventing—just make sure a human gets the credit.