December 11, 2025

Top Section 101 Patent Eligibility Stories of 2025

Holland & Knight Section 101 Blog
Anthony J. Fuga
Section 101 Blog

Top off the eggnog, straighten the twinkle lights and clear a space between the nutcrackers and the model train set for a fresh stack of Top Section 101 Stories of the Year. It's our annual sleigh ride through Section 101. Nothing says "peace on earth" like debating mental processes over panettone.

Holiday Movies. This year's rotation gets a twist: "Home Alone 2" is back in the mix, along with "The Muppet Christmas Carol." "National Lampoon's Christmas Vacation," "Feast of the Seven Fishes," "The Santa Clause" and "The Holiday" are in no danger of being demoted. And "The Nightmare Before Christmas" has been on repeat for about two months now. I'll blame a child for that one.

"Die Hard"? Still yes. "Scrooged"? Still on the naughty list, along with "A Christmas Story," which remains an immediate veto. "The Polar Express"? Oh, no – what a travesty.

Holiday Table. The panettone graduates from countertop snack to breakfast strategy. St. Lucia buns show up all week, and the glögg is now a two‑spice operation (cardamom and a single clove). As for the "Feast of the Seven Fishes," we're evolving, but still no shellfish.

Now, onto the list.

No. 3: Congress Dusts Off PERA; the Chorus for "Clarity" Gets Louder

The U.S. Senate Committee on the Judiciary returned to the Patent Eligibility Restoration Act (PERA) this fall, with an October hearing reprising familiar refrain. Supporters emphasized the costs of uncertainty and its drag on investment, particularly in diagnostics and artificial intelligence (AI). Critics warned the bill could swing the door too far open, flooding the system with low-quality patents and pushing costs downstream. Veteran Federal Circuit voices signaled what most practitioners already know (or at least expect): Lasting change will require Congress, not the courts.

What matters in practice is less the headlines and more the contours. PERA continues to grapple with a carveout for processes "substantially economic, financial, business, social, cultural, or artistic," balanced against language that could preserve eligibility where claims cannot practically be performed without a machine.

How those phrases ultimately resolve will define whether long‑controversial classes of software and business‑method claims find new life or remain fenced out. For now, prudent portfolio planning means scenario‑mapping the families that would benefit under a broadened regime and those that will need to live with the status quo.

No. 2: The USPTO Recalibrates 101 – Desjardins, "Tie Goes to the Runner" and Examiner Guidance

Inside the U.S. Patent and Trademark Office (USPTO), 2025 was the year Section 101 moved decisively toward being a true gatekeeper. In August, the Deputy Commissioner for Patents issued a memorandum reiterating core points that matter in real prosecutions, especially for AI and machine learning (ML):

  • "Mental Process" Means Mental Process. If a limitation cannot practically be performed in the human mind, it is probably not a mental step.
  • Distinguish "Recites" from "Involves." "Examiners should be careful to distinguish claims that recite an exception (which require further eligibility analysis) from claims that merely involve an exception."
  • Analyze the Claim as a Whole. Improvements to computer functionality or another technology integrate an exception into a practical application.
  • Close Calls Are Not Rejections. The Office underscored the preponderance standard: Examiners should reject under Section 101 only when ineligibility is more likely than not.

Then came the Appeals Review Panel decision in Ex parte Desjardins, vacating a sua sponte Patent Trial and Appeal Board (PTAB) Section 101 rejection of a continual‑learning invention and emphasizing that software can effect non‑abstract improvements to computer technology. The decision leaned into Enfish and reminded the examining corps that Sections 102, 103 and 112 (not Section 101) are the "traditional and appropriate tools" to police scope.

December brought more, and the message was straightforward: When prepared correctly, declarations can supply objective evidence for Step 2 considerations, including technological improvements, "significantly more" and how alleged mental‑process characterizations miss what is technically happening.

For practitioners, this combination – Desjardins plus the August memo plus Subject Matter Eligibility Declaration (SMEDs) – creates a workable pathway for AI claims that read like technology, not ends‑oriented automation.

No. 1: Recentive Takes Center Stage – and the Supreme Court Sits It Out

The U.S. Court of Appeals for the Federal Circuit's decision in Recentive Analytics v. Fox was the year's headline for many. Framed as a question of first impression, 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 under Section 101.

First, the court rejected the idea that iterative training, dynamic updates or real‑time adjustments – features "incident to the very nature of machine learning" – convert functional, ends‑oriented claims into eligible technology. Without concrete technical detail on how the ML model or infrastructure is improved, claims that recite "using machine learning to optimize _____" remain abstract.

Second, the court faulted the lack of a "how." Functional claiming that states the result without the claimed mechanism invites invalidation. The opinion's closing caveat matters: Machine learning "may lead to patent‑eligible improvements," but not where the innovation is simply applying machine learning in a new domain.

The U.S. Supreme Court denied Recentive's petition this month, declining to revisit eligibility for machine learning claims. That leaves in place a stark, but useful, juxtaposition. At the Federal Circuit, it is not enough that AI replaces human judgment in a new field; claims need to capture architecture, training methodology, data structures, constraints and other concrete mechanisms that change how the system operates. At the USPTO, however, the tone and training have shifted toward recognizing those same mechanisms as technological improvements at Step 2A, with "close call" rejections discouraged and evidentiary declarations welcomed.

The result is a familiar split: more applicant‑friendly pathways during examination and stiffer headwinds in litigation for purely functional AI claims.

What to Do with This in 2026

Draft like a technologist. Claims that survive Section 101 are the ones that read like engineering, not aspiration. For AI, that typically means reciting model architecture choices, training regimes that change how the model learns, compute‑ or data‑path constraints that force different operation, or data structures that enable capabilities not practically doable by a human.

Use the record. The August memo's "tie goes to the applicant" standard, the AI examples and SMEDs create levers that should work at the USPTO.

But plan for the forum split: What persuades an examiner at Step 2A needs to be sturdy enough to withstand a defendant's "ends‑oriented" narrative at Rule 12.

Did I miss a top story or commit a holiday‑movie heresy? Let me know.

Have a great holiday season, and I'll see you in 2026.

Related Insights