USPTO Issues Guidelines for Inventions Created with AI

From articles in our news feeds to conversations with friends, the topic of Artificial Intelligence (AI) seems inescapable. This trend has even pervaded the world of intellectual property, where attorneys and inventors are increasingly asking: who owns the rights to products created by ChatGPT or Gemini? Recently, the United States Patent and Trademark Office (USPTO) published guidelines to help attorneys and inventors better understand the role of AI in creating patentable material. As the USPTO itself states, “This guidance…does not have the force and effect of law.” Instead, the guidelines clarify how the agency will handle questions about artificial intelligence based on rulings from the United States Supreme Court and the Federal Circuit.

Importantly, the guidelines stated that AI cannot serve as a named inventor or co-inventor on patents. To better understand why ChatGPT cannot be named as an inventor, it might make sense to look at how the Supreme Court understood the act of invention in Pfaff v. Wells Electronics, Inc. Here, the Supreme Court determined that “invention,” as the word is used in the Patent Act, refers to an inventor’s conception rather than their actual product. In other words, when Thomas Edison came up with the idea for the light bulb, that was the moment of invention, as opposed to when the light bulb was manufactured. And since AI lacks a mental capacity, it cannot conceive, making it unable to meet the standard required for invention.

Alright, then, you might think: AI cannot be named as an inventor, but how do I ensure that I’ve contributed enough to an AI-assisted project to be considered an inventor? Here, the USPTO guidelines direct individuals to the Pannu factors, criteria that the courts use to determine whether a natural person qualifies as an inventor. According to Pannu, someone reaches the threshold of an inventor if they: 

“(1) contribute in some significant manner to the conception or reduction to practice of the invention,

(2) make a contribution to the claimed invention that is not insignificant in quality, when that contribution is measured against the dimension of the full invention, and 

(3) do more than merely explain to the real inventors well-known concepts and/or the current state of the art.”

So, as long as someone meets the Pannu factors, they can be considered an inventor on a utility patent application, even if they have AI assistance. And according to the USPTO guidelines, the agency also uses the Pannu factors to determine whether someone qualifies as an inventor on plant and design patents.

It’s important to understand that identifying a problem and asking AI to solve it do not make someone an inventor. For example, if I queried ChatGPT to “create a new medicine to cure breast cancer,” and by some miracle, it came up with an entirely novel solution, I did not work on the medicine enough to meet the Pannu factors. However, the USPTO guidelines note that if AI provides an answer and an individual builds upon it, that person might reach the status of inventor. So maybe ChatGPT suggested a couple of compounds that would be helpful in my cancer-curing medicine. Then, I tested out those ingredients and combined them with others, arriving at a new drug formula. Since I’ve contributed to the drug’s creation, I might now qualify as an inventor even with the initial AI assistance.

These recent guidelines from the USPTO provide a useful framework for attorneys and inventors alike to understand the role of AI in invention. As AI becomes increasingly integrated with our professional lives, more cases related to its role in patent creation are sure to arise. Keep following the IdeasESQ Blog at Schott, P.C. for more content concerning the role of AI in intellectual property law.