On November 28, the USPTO did something we don’t see often, it completely walked back guidance it issued less than a year ago. The Patent and Trademark Office rescinded its February 2024 inventorship guidance for AI-assisted inventions and replaced it with a cleaner, more practical framework.

The new guidance confirms what most practitioners already understood intuitively: the Pannu factors were never meant for AI analysis in the first place.

Why the Change Matters

“The Pannu factors only apply when determining whether multiple natural persons qualify as joint inventors. Pannu is inapplicable when only one natural person is involved in developing an invention with AI assistance because AI systems are not persons and therefore cannot be ‘joint inventors’ so there is no joint inventorship question to analyze.”

The bottom line: “The same legal standard for determining inventorship applies to all inventions, regardless of whether AI systems were used in the inventive process. There is no separate or modified standard for AI-assisted inventions.”

Quick Refresher: The Pannu Factors

For context, Pannu v. Iolab Corp. established a three-part test that courts use when people dispute who should be named as inventors on a patent. Each claimed joint inventor must:

• Contribute to conception or reduction to practice – Not just execute someone else’s idea, but actually contribute to forming the invention itself.
• Make a non-trivial contribution – The contribution has to matter when you look at the full scope of what’s claimed.
• Do more than explain known concepts – Teaching existing knowledge doesn’t make you an inventor.

This test helps sort out legitimate co-inventors from people who merely assisted. 

The principle goes back further than Pannu, though. In 1853, the Supreme Court noted in O’Reilly v. Morse that “it can make no difference…whether [the inventor] derives his information from books, or from conversation with men skilled in the science.” Getting help doesn’t dilute inventorship. The USPTO quotes this exact language in the new guidance to explain why AI assistance works the same way.

But here’s the key: Pannu presupposes that all parties in the analysis are natural persons who could theoretically hold patent rights. When one of the “parties” is a software tool, the framework breaks down. You’re not resolving a dispute between competing inventors, you’re just asking whether a human being conceived the invention.

What This Means in Practice: Don’t List AI as an Inventor

This should go without saying, but the USPTO makes it explicit anyway. The Office presumes that inventors named on the application data sheet or oath/declaration are the actual inventors. But if you list an AI system, expect “a rejection under 35 U.S.C. 101 and 115, or other appropriate action…for all claims in any application that lists an AI system or other non-natural person as an inventor or joint inventor.”

Every claim. Not just some of them.

What You Need to Remember

• Apply the same inventorship standards you always have. There’s no special analysis for AI-assisted inventions.
• Never name AI systems as inventors. They’re tools, not legal persons. This should be obvious, but apparently it needs to be said.
• Focus on whether your client can describe the invention with sufficient particularity. That’s the conception test, same as always.
• Watch your foreign priority claims. Make sure you have at least one natural person inventor in common between applications.
• Document your process. When AI assists in development, keep records that show human conception. You’ll thank yourself later if inventorship ever becomes an issue.
• Counsel your clients proactively. Explain that using AI tools doesn’t dilute their inventorship rights. They’re using assistance, not sharing credit.

Why This Matters Beyond Inventorship

The USPTO’s reversal signals more than a procedural fix. It reflects a maturing view of AI as a routine part of invention rather than a category requiring special legal treatment. By discarding the 2024 framework and reaffirming traditional conception as the touchstone, the Office is aligning with judicial precedent, prioritizing clarity, and emphasizing that AI is an instrument used by human inventors, not a legal actor and not a basis for a separate inventorship doctrine. This approach normalizes AI-assisted innovation, places responsibility on practitioners to document human conception, and provides a clean, scalable standard that can accommodate rapidly evolving technologies. And while this guidance settles the question of inventorship for now, it also sets the stage for broader, ongoing conversations about how AI will reshape patent practice in the years ahead.

Quick Answers

Can AI be listed as an inventor?
No. Only natural persons can be inventors. Applications listing AI systems will be rejected under 35 U.S.C. §§ 101 and 115.

Do different rules apply to AI-assisted inventions?
No. The same inventorship standard applies regardless of whether AI was used in the process.

Can I claim priority to a foreign application naming AI as an inventor?
No. Priority claims require at least one natural person inventor in common. If the foreign application names only AI, the claim won’t work. If it names both humans and AI, your U.S. application must list only the human inventors (with at least one in common).

How does inventorship work when multiple people use AI tools?
Apply the Pannu factors to each human contributor. The use of AI doesn’t change how you evaluate inventorship among people.

When did this take effect?
November 28, 2025. It applies immediately to pending applications and new filings.