Patent Eligibility Wars: A New Hope at the PTAB

Key Updates

• Patent applicants challenging subject-matter eligibility rejections under 35 U.S.C. § 101 are seeing improved results at the PTAB. Recent data show that the PTAB reversal rate rose from about 9.4% to about 21% during Q4 2025, before moderating to about 15% from April through July 2026.
• The increase follows the USPTO’s precedential decision in Ex parte Desjardins, which emphasized that software and artificial intelligence inventions may provide patent-eligible technological improvements even when the improvement resides in logical structures, data processing or machine-learning techniques.

Why It Matters

• Recent PTAB reversals reveal recurring persuasive arguments, including: challenging an examiner’s overgeneralized characterization of the claims; requiring the examiner to precisely identify which limitations allegedly constitute the judicial exception; demonstrating a concrete technological improvement under step 2A, prong two; and challenging unsupported assertions that additional elements are merely well-understood, routine and conventional.
• For software and artificial intelligence applicants, the recent decisions reinforce the value of developing the technological-improvement story during drafting and prosecution by identifying the technical problem, the claimed mechanism that addresses it and the resulting improvement in computer operation or another technical field.

Next Steps

• Patent applicants and prosecution teams should review software, artificial intelligence and data-driven applications with pending or anticipated §101 issues to determine whether the claims and prosecution record preserve the strongest path for appeal.

For much of the last decade, applicants appealing § 101 rejections to the Patent Trial and Appeal Board (PTAB) could be forgiven for feeling that the odds were far, far away from being in their favor. Historical studies routinely placed the PTAB’s affirmance rate for examiner eligibility rejections near 90%.

That pattern began to change in September 2025. One of the earliest and most visible signals came in Ex parte Desjardins, where an appeals review panel reversed a § 101 rejection involving techniques for training machine-learning models. The panel emphasized that the claims had to be evaluated as a whole and relied on Enfish for the proposition that improvements in computer technology need not take the form of new physical components. Instead, technological advances may reside in logical structures and processes. The U.S. Patent and Trademark Office (USPTO) designated Desjardins precedential in November 2025 and subsequently incorporated its reasoning into eligibility guidance.

The Data: A Partial Retreat, But Still Above the Old Baseline

The early statistical response was substantial. The PTAB’s reversal rate for examiner § 101 rejections increased from approximately 9.4% to approximately 21% during the initial period following September 2025. The rate later moderated to approximately 15% for April through July 2026. The recent data therefore suggest an initial surge followed by moderation — not a return to the previous baseline. The latest published ten-month view is summarized below:

The current picture therefore looks less like a wholesale abandonment of § 101 and more like a meaningful change in the PTAB’s willingness to police the quality of eligibility rejections. The key question for applicants is what distinguishes the appeals that are winning.

The statistics are only part of the story. The PTAB’s 2026 reversal decisions show recurring analytical shortcomings that applicants are successfully exploiting, and those decisions provide a practical roadmap for both prosecution and appeal strategy.

Practical Takeaways from PTAB Trends

1. Challenge an overly generalized or claim-inconsistent abstract idea. An application’s possible commercial use does not necessarily make the claim a commercial interaction. Compare the asserted judicial exception against the claim language and challenge characterizations that depend on unclaimed purpose, context or embodiments.

Example: In Ex parte Shopify Inc., Appeal No. 2026-000163, the examiner characterized claims involving an augmented-reality interface as directed to commercial interactions involving purchase history and post-purchase activities. The applicant pointed out that the claims did not recite purchases, advertising, marketing or sales activities; those concepts came from examples in the specification rather than the claims themselves. The PTAB agreed that the examiner had improperly imported purchase-related concepts to establish a “commercial or legal interactions” abstract idea and reversed the § 101 rejection.

2. Require the examiner to identify the judicial exception with precision. Force a limitation-by-limitation Step 2A analysis. Require the examiner to identify which limitations allegedly constitute the judicial exception and which remain as additional elements. If the rejection does not make that distinction clear, make the deficiency explicit before addressing step 2A, prong two or step 2B on the merits.

Example: In Ex parte Predict HQ Limited, Appeal No. 2025-003791, the PTAB initially affirmed the § 101 rejection but changed course on rehearing in July 2026. The applicant argued that the examiner had not adequately distinguished the limitations allegedly constituting the judicial exception from the “additional elements” that had to be evaluated under step 2A, prong two and step 2B. The PTAB agreed that the examiner had not parsed the claim with sufficient specificity. Without a clear delineation, the applicant could not meaningfully argue that particular limitations integrated the alleged exception into a practical application or otherwise conferred eligibility.

3. Tie the claimed mechanism to a specific technological improvement. Build the technological-improvement story into the specification and connect that improvement to actual claim limitations. Describe the technical problem, identify the particular architecture or process that addresses it and explain concrete effects such as reduced memory usage, faster processing, improved reliability or a different manner of operating the computer. Where appropriate, amend the claims so the technical mechanism producing the improvement is express, and focus the eligibility argument on step 2A, prong two.

Example: In Ex parte Oracle International Corporation, Appeal No. 2026-000106, the claims concerned techniques for iterating through modified graph data using an iterator stored in RAM. The applicant identified a concrete improvement: the claimed RAM-based graph iterator accelerated graph traversal and maintained query performance in a system using delta logging. Importantly, the specification described how the architecture improved query efficiency and avoided special indices or special-purpose hardware registers. The PTAB concluded that the claimed configuration improved computer functionality in the sense contemplated by Enfish and reversed the § 101 rejection.

4. Improvements to machine-learning models can themselves be technological improvements. For AI and machine-learning inventions, identify a particular improvement to model operation, training, accuracy, structure or efficiency, and ensure that improvement is both supported by the specification and reflected in the claims. A generic assertion that AI is involved is less persuasive than a concrete explanation of how the claimed model or training architecture operates differently or better.

Example: In Ex parte Surescripts, LLC, Appeal No. 2026-001183, the claims involved augmenting medication-history data using a specifically trained neural-network model. The PTAB agreed that portions of the claims implicated abstract concepts, but under Step 2A, Prong Two the applicant focused on the particular way the neural network was trained and updated. The specification described how that training architecture improved incomplete, mismatched or poor-quality data and could improve the accuracy of the model itself. Relying on Desjardins, the PTAB concluded that the claimed improvement in training the neural-network model constituted an improvement to computer technology and integrated the judicial exception into a practical application.

5. Hold examiners to the evidentiary requirements of step 2B. Challenge unsupported conventionality findings. If the examiner reaches step 2B, require factual support for assertions that the relevant additional elements are well-understood, routine and conventional; a conclusory characterization should not substitute for the evidentiary showing required by USPTO guidance.

Example: In Ex parte Martineau, Appeal No. 2026-000138, the PTAB reversed a § 101 rejection from Technology Center 3600 after finding that the examiner had not adequately applied the USPTO framework. Among other issues, the examiner relied on assertions that additional claim elements were well-understood, routine and conventional without supplying the factual support required by USPTO guidance. The PTAB concluded that the rejection did not adequately establish ineligibility.

What These Trends Mean for Patent Applicants

The recent PTAB decisions do not eliminate the uncertainty surrounding § 101, and the decline from the initial reversal surge suggests that applicants should not expect every eligibility rejection to meet an untimely end. But the available data and 2026 decisions point in the same general direction: examiner eligibility rejections are surviving appeal less frequently than before fall 2025, and the PTAB is increasingly demanding a disciplined analysis of the actual claim language, the identified judicial exception and any claimed technological improvement.

For software, artificial intelligence and other data-driven inventions, the takeaway is not that § 101 has disappeared. Rather, applicants may finally have a new hope: a better-developed framework for challenging overbroad eligibility rejections and a PTAB that appears more willing to enforce that framework on appeal.

For more information regarding subject-matter eligibility, software and artificial intelligence patent strategy, or appeals before the PTAB, contact Elias Larson or your preferred Polsinelli attorney.