AI-Aided Invention Questions Swirl After Patent Agency Guidance

Feb. 15, 2024, 10:04 AM UTC

New patent office guidance on how much human involvement is necessary for AI-assisted inventions to win protection highlights thorny questions for attorneys drafting patent applications.

The US Patent and Trademark Office said AI-assisted inventions can be eligible for a patent if a human was significantly involved in developing it. The agency previously said that AI itself can’t be an inventor deserving patent protection, in a decision backed by the US Court of Appeals for the Federal Circuit.

Attorneys will have to work to ensure they can demonstrate enough human involvement in an AI-assisted invention to persuade the agency to issue a patent that’s strong enough to survive any challenges.

“We should be asking our clients, ‘was AI involved?’ And if we don’t do that it could be a violation of a lawyer’s duty of disclosure at the PTO,” Michael Word, a Chicago-based member at Dykema Gossett PLLC said. “If you put your head in the sand, you might not be meeting your requirement.”

Simply recognizing a problem and posing it as a prompt to an AI system isn’t a significant human contribution under the guidance, said Erik A. Huestis, co-chair of Foley Hoag LLP’s technology industry group.

“There’s a gray area. Where those lines are going to be drawn is going to be a significant source of case law in the coming years,” Huestis said. “It highlights the need to really document and understand” how AI tools are used, “even to the level of putting in your lab notebook what prompt did you provide.”

Companies should document human involvement in AI-assisted inventions before the patent application process to help show that inventions are patent-eligible, Manita Rawat, a partner at Morgan, Lewis & Bockius LLP, said.

“We now have AI technology where some of these inventions are coming up with reiterations on their own,” Rawat said.

Litigation Factor

Questions around how much credit for an AI-assisted invention a human can claim will also bubble up in litigation, attorneys said.

“Accused infringers will take a very hard look at what the AI was doing, versus the humans, and challenging whether that was enough,” David Lisson, a partner at Davis Polk & Wardwell LLP, said.

AI prompt logs will come into play in litigation, Huestis said.

“I look forward to reading transcripts where somebody is being asked, ‘OK, exactly what did you feed to the generative AI model in order to have it spit out this diagram, this molecule?’” Huestis said.

There could be opportunities for litigators to find vulnerabilities in patents’ validity during discovery if an AI system contributed to a claimed invention and that contribution wasn’t flagged to the patent office, Word said.

“I’m certainly going to add this to my list of discovery questions when defending patent cases,” Word said.

To contact the reporters on this story: Lauren Castle in Dallas at lcastle@bloombergindustry.com; Michael Shapiro at mshapiro@ic.bloombergindustry.com

To contact the editors responsible for this story: Keith Perine at kperine@bloomberglaw.com; Adam M. Taylor at ataylor@bloombergindustry.com

Learn more about Bloomberg Law or Log In to keep reading:

See Breaking News in Context

Bloomberg Law provides trusted coverage of current events enhanced with legal analysis.

Already a subscriber?

Log in to keep reading or access research tools and resources.