USPTO Director Reverses PTAB § 101 Rejection, Reaffirms Patent Eligibility for AI Innovations
On September 26, 2025, John Squires, Under Secretary of Commerce and Director of the USPTO, took the unusual step of convening an Appeals Review Panel (ARP) and vacating a Patent Trial and Appeal Board (PTAB) decision that had rejected an artificial intelligence-related invention being ineligible for patenting under 35 U.S.C. §101[1]. The Director’s decision sends an unequivocal signal that categorical exclusions of AI innovations threaten America's technological leadership and that Examiners must faithfully apply established Federal Circuit precedent, particularly Enfish, when evaluating patent eligibility. This decision, coming just four days after Director Squires took office on September 22, 2025, demonstrates his commitment to ensuring that genuine technological improvements receive appropriate patent protection.
“Categorically excluding AI innovations from patent protection in the United States jeopardizes America's leadership in this critical emerging technology… Examiners and panels should not evaluate claims at such a high level of generality.” [2]
For companies developing AI and machine learning technologies, this reversal signals a more favorable environment for patent eligibility.
Background: Patent Eligibility and its Difficulties for Software and AI Companies
Patent applicants in artificial intelligence and software have long encountered steep challenges at the United States Patent and Trademark Office (USPTO). Too often, rejections hinge not on the novelty or inventiveness of the technology itself, but on whether the claims are said to fall within the judicially created category of “abstract ideas.” [3] Alongside the statutory requirements of novelty, nonobviousness, and adequate written description, an invention must also qualify as patent-eligible subject matter under §101. Even groundbreaking advances can be deemed ineligible if they are characterized as a law of nature, natural phenomenon, or abstract idea—collectively referred to as “judicial exceptions.”[4] The abstract idea category has proven especially murky, encompassing mathematical concepts[5], certain methods of organizing human activity[6], and so-called “mental processes.” [7] For software inventions, the last of these is particularly troublesome, as examiners sometimes argue that program steps could theoretically be carried out by a human with pen and paper, regardless of the impracticality. The uneven application of these standards has produced a patchwork patent landscape, where similar innovations receive drastically different treatment depending on the examiner, complicating patent strategy and driving up uncertainty in cost and outcome.
To evaluate whether a claim falls within a judicial exception, patent examiners apply a two-step inquiry (often framed as Step 2A, prong one and prong two of the Alice/Mayo analysis). In the first prong, the examiner asks whether the claim recites an abstract idea, a law of nature, or a natural phenomenon. If the answer is yes, the inquiry moves to the second prong, which examines whether the claim as a whole incorporates the identified exception into a practical application. Where the claim integrates the exception in this way, it is considered patent-eligible under §101, though it must still satisfy the other requirements of patentability.
This enforcement challenge became particularly acute following the August 4, 2025, memorandum from Deputy Commissioner Charles Kim[8], which reminded examiners that §101 rejections should only be made when ineligibility is “more likely than not”—establishing a concrete preponderance of evidence standard that creates a presumption in favor of eligibility for borderline cases. Despite this guidance emphasizing restraint in §101 rejections, the PTAB panel in Desjardins proceeded to enter a sua sponte §101 rejection that directly contradicted these principles.
PTAB’s Overbroad Reasoning and Director Squires’ Intervention
DeepMind Technologies appealed a rejection over prior art of their application relating to continual learning in machine learning models. The PTAB panel affirmed the obviousness rejection, but also entered a new ground of rejection under §101, essentially equating any machine learning with an unpatentable ‘algorithm’ and dismissing remaining claim elements as ‘generic computer components’ without adequate explanation—precisely the type of categorical exclusion that the August 4th memorandum cautioned against. The Board dismissed the application’s improvements in efficiency and continual learning as insufficient to confer eligibility under prong two.
The Director vacated the PTAB’s rejection, criticizing the panel for equating all machine learning with unpatentable “algorithms.” He emphasized that:
AI claims should not be analyzed at a “high level of generality.”
Improvements to continual learning and model efficiency constitute technological improvements, not abstract ideas.
Panels must faithfully apply Enfish[9] and related Federal Circuit precedent recognizing software as capable of non-abstract improvements.
This does not represent a complete win for DeepMind. The Director noted that rejections under §102 (novelty), §103 (obviousness), and §112 (written description and enablement) are the proper tools for limiting the scope of patent protection, but stressed that §101 rejections should be used rarely and only where appropriate.
Nevertheless, Director Squires delivered particularly sharp criticism of the panel’s approach, stating that their “sua sponte action is most troubling, as it eschewed the clear teachings of Enfish, and instead substituted only a cursory analysis that ignored this well-settled precedent.” The Director emphasized that panels “should treat such precedent with more care, especially when acting sua sponte,” highlighting the procedural impropriety of entering new rejections without proper legal foundation.[10]
Reinforcing the August 4th Memorandum
The Desjardins decision directly reinforces the principles articulated in the August 4, 2025, memorandum from Deputy Commissioner Charles Kim. That memorandum established that examiners must demonstrate a greater than 50% probability that claims are ineligible before making §101 rejections and specifically cautioned against categorizing AI limitations that “cannot practically be performed in the human mind” as mental processes. The memorandum also emphasized analyzing claims “as a whole” rather than evaluating elements “in a vacuum, completely separate from the recited judicial exception.” The PTAB panel's approach in Desjardins violated each of these principles by: (1) failing to meet the preponderance standard for ineligibility; (2) improperly categorizing complex machine learning operations as abstract algorithms; and (3) analyzing claim elements in isolation rather than considering their integrated technological contribution.
Conclusion and Next Steps
For AI CTOs and in-house counsel, this decision provides critical reassurance:
Core innovations in training strategies, efficiency, and machine learning now stand on firmer §101 ground.
Patent applications improving how machine learning models operate should be viewed as technological improvements, not abstract ideas.
Broad, categorical §101 rejections against AI, and indeed all software innovations, should face greater scrutiny going forward.
Arguments against a patent ineligibility challenge should reference both the August 4th memorandum's preponderance standard and the Director's criticism of categorical AI exclusions in Desjardins.
Counsel should re-evaluate pending cases with §101 rejections and consider whether an invention’s technical improvements to efficiency, storage reduction, and processing requirements provide Enfish-style eligibility. Counsel should also consider reviving applications or filing continuations of applications previously abandoned after §101 rejections. For new applications, the specification should emphasize particular technical improvements to operations, and the elements or functions that provide these improvements should be reflected in the claims.
This decision represents a strong policy signal from the highest levels of the USPTO. AI innovators should work closely with counsel to ensure claims highlight technical improvements and practical applications, providing the strongest possible footing under §101.
This decision represents more than policy guidance. It constitutes a direct mandate from the highest USPTO authority that § 101 rejections must be grounded in proper legal analysis rather than categorical exclusions. When combined with the August 4th memorandum's preponderance standard, applicants now have unprecedented ammunition to challenge overbroad eligibility rejections. The Director's willingness to personally author this decision within his first week signals that §101 reform will be a priority throughout his tenure, creating the most favorable examination environment for AI and software innovations in over a decade.
For questions regarding how the Director’s decision may affect your pending applications or portfolio strategy, please contact partners Daniel Rose, Intellectual Property Department at daniel.rose@pierferd.com; or Sid Kapoor, Intellectual Property Department, at sid.kapoor@pierferd.com.
This publication and/or any linked publications herein do not constitute legal, accounting, or other professional advice or opinions on specific facts or matters and, accordingly, the author(s) and PierFerd assume no liability whatsoever in connection with its use. Pursuant to applicable rules of professional conduct, this publication may constitute Attorney Advertising. © 2025 Pierson Ferdinand LLP.
[1] Appeal 2024-000567, Application 16/319,040 (PTAB Apps. Rev. Sept. 26, 2025) (“Desjardins”).
[2] Id. at 9.
[3] “USPTO Issues New Guidance on Patent Subject Matter Eligibility under 35 U.S.C. § 101”, Rose et al., Pierson Ferdinand (August 28, 2025) (available at: https://pierferd.com/insights/uspto-issues-new-guidance-on-patent-subject-matter-eligibility).
[4] Diamond v. Chakrabarty, 447 U.S. 303, 309 (1980).
[5] Gottschalk v. Benson, 409 U.S. 63, 71-72 (1972).
[6] Alice Corp. v. CLS Bank Int’l, 573 U.S. 208, 220 (2014) (“[T]he concept of intermediated settlement is ‘a fundamental economic practice long prevalent in our system of commerce,’ and thus an abstract idea (and, like hedging, a method of organizing human activity).”).
[7] Mayo Collaborative Servs. v. Prometheus Labs. Inc., 566 U.S. 66, 71 (2012) (“‘[M]ental processes[] and abstract intellectual concepts are not patentable, as they are the basic tools of scientific and technological work’” (quoting Benson, 409 U.S. at 67)).
[8] “Reminders on evaluating subject matter eligibility of claims under 35 U.S.C. 101,” Deputy Commissioner Charles Kim, USPTO (August 4, 2025) (available at: https://www.uspto.gov/sites/default/files/documents/memo-101-20250804.pdf).
[9] Enfish, LLC v. Microsoft Corp., 822 F.3d 1327, 1335 (Fed. Cir. 2016).
[10] Appeal 2024-000567 at 9-10.
