Haynes Boone Partner Dina Blikshteyn spoke with MLex as a recent decision in Recentive Analytics v. Fox Corp. is generating heightened interest from patent litigants.
In support, The Nielsen Company LLC points to inclusion in the specification of one of the patents of 12 paragraphs of “detailed example implementations of the feature map network and the probability map network” and says that, unlike the claims at issue in Recentive, the ‘588 and ‘782 patents “use a specific neural network architecture to improve image processing, a technical issue that requires a technical solution.”
That’s the type of narrative Blikshteyn says AI innovators should be working to develop long before their innovation is challenged in court.
Blikshteyn told MLex she is seeing “a huge uptick in patent eligibility rejections in the AI space.”
It’s a trend Blikshteyn said she expects will continue in the wake of Recentive.
That’s why she advises clients to approach patent prosecution “almost the same way” as they would litigation or post-grant proceedings.
“You need a story, and you have to include that story up front. Try to predict ways of minimizing a patent eligibility rejection at the time of drafting — rather than at the time the patent office picks up and decides to examine the patent application,” Blikshteyn suggests.
Additionally, a deep dive on different AI models can be beneficial when drilling down on a claimed technical improvement. In industries like pharmaceuticals and diagnostics, where AI training data is regarded as the “crown jewels” of the claimed advance, Blikshteyn notes applicants may “need to be more creative” during drafting.
“Start looking at the combinations of different AI models and try to find a technical improvement that way,” she adds.
“As the Federal Circuit recently held, ‘patents that do no more than claim the application of generic machine learning to new data environments, without disclosing improvements to the machine learning models to be applied, are patent ineligible under § 101.’ Recentive Analytics, Inc. v. Fox Corp. 134 F.4th 1205, 1216 (Fed. Cir.2025). Yet, that is precisely what the ‘041 Patent claims — i.e., using generic machine learning technology as a tool to carry out longstanding mental methods of analyzing stained tissue slides,” the motion adds.
The flurry of activity is unsurprising to Blikshteyn.
She said she followed Recentive’s case closely and had hoped the patent owner would not appeal the dismissal by Williams in order “to avoid exactly this outcome” at the Federal Circuit.“ This is not a great decision for AI patents and AI patent applications,” she predicted.
To read the full article from MLex, click here.