Federal Circuit hands win to Fox in machine learning patent infringement dispute

The patents-in-suit include:

• U.S. Patent No. 10,911,811 (“Systems and methods for automatically and dynamically generating a network map”)
• U.S. Patent No. 10,958,957 (“Systems and methods for automatically and dynamically generating a network map”)
• U.S. Patent No. 11,386,367 (“Systems and methods for determining event schedules”)
• U.S. Patent No. 11,537,960 (“Systems and methods for determining event schedules”)

In its appeal against Fox’s motion to dismiss, Recentive had argued that “the concept of preparing network maps [had] existed for a long time” and that prior to computers “networks were preparing these network maps with human beings.”

While the plaintiff admitted that its patents “do not claim the machine learning technique itself” but rather “claim the application of the machine learning technique to the specific context[s] of event scheduling and network map creation”, it argued that its patents were eligible because they involve:

“The unique application of machine learning to generate customized algorithms, based on training the machine learning model, that can then be used to automatically create…event schedules that are updated in real-time.”

Recentive further described its patents as introducing “the application of machine learning models to the unsophisticated, and equally niche, prior art field of generating network maps for broadcasting live events and live event schedules.”

This is the Federal Circuit’s opinion:

To reach its conclusion, the court used the Alice Corporation v. CLS Bank International two-step test:

1. Step 1 (determine whether the claims at issue are directed to one of those patent-ineligible concepts): The court said the claims of the patents failed as they were “directed to the abstract ideas of producing network maps and event schedules, respectively, using known generic mathematical techniques”. Recentive’s argument that its patents are eligible simply because they introduce machine learning techniques to the fields of event planning and creating network maps directly conflicts with our § 101 jurisprudence, it added.
2. Step 2 (consider the elements of the claim both individually and ‘as an ordered combination’ to determine whether the additional elements ‘transform the nature of the claim’ into a patent eligible application): The claims failed to show an “inventive concept” as “the machine learning limitations were no more than broad, functionally described, well-known techniques” and claimed “only generic and conventional computing devices,” the court found, adding:

“We perceive nothing in the claims that would transform the Machine Learning Training and Network Map patents into something ‘significantly more’ than the abstract idea of generating event schedules and network maps through the application of machine learning,” the court opined.

Finally, the court rejected Recentive’s argument that the district court should have granted it leave to amend. This is because the appellant failed to propose any amendments or identify any factual issues that would alter the § 101 analysis, it held.

‘Problematic for future AI-related patents’

Some members of the community have already taken to LinkedIn to react to the court’s opinion, including Ropes & Gray partner Matt Rizzolo. He wrote that while the court held that claims that do no more than apply established methods of machine learning to a new data environment are not patent eligible, it also wrote that machine learning is a “burgeoning and increasingly important field and may lead to patent eligible improvements in technology”. The key, he highlighted is that claiming the application of generic machine learning to new environments, without disclosing improvements to the machine learning models themselves, does not pass muster under 101.

Mr. Rizzolo’s post sparked some conversation in the comments section. 

Doug Pittman, Founder of BoardActive and BrandDrop, warned that “AI is nothing more than algorithms scraping the internet and gathering the requested info on what others own, especially copyrighted and other IP, regrouping the data stolen from others and paragraphing it into words for smart usage”.

Disagreeing with Mr. Pittman, Thomas H. Rousse, a former Law & Science Fellow at Northwestern University School of Communication, wrote that “patent owners own the process or method described, not the language or information in the patent” and that while plenty of the outputs/uses are debatable at best, patents “should certainly be available for ‘scraping’”.

Matthew Dowd, Founder and Principal at Dowd Scheffel PLLC, added that this decision is likely to be “very, very problematic for many, many AI-related patent applications and issued patents that have been filed/issued over the past decade”. There are currently over 78,000 patent/application families with “machine learning” in the claims in the USPTO database.

Counsel

A team at Goodwin Procter LLP represented Recentive Analytics: Robert Frederickson III, Jesse Lempel, and Alexandra D. Valenti.

Meanwhile, Fox was represented by Pillsbury Winthrop Shaw Pittman LLP’s Ranjini Acharya, Michael Zeliger, Evan Finkel, and Michael Shigeyori Horikawa.

In a statement, Pillsbury Winthrop Shaw Pittman LLP said that the decision “affirms [their] longstanding contention that technological improvements created merely by the ordinary use of generic machine language (or AI) are not patent eligible.” The firm added that the court’s ruling “has far-reaching implications, as courts and companies grapple with the best way to commercialize and use new AI technologies across a wide range of fields.”