Patent Office Appeals Review Decision Suggests Policy Shift Toward Broader Eligibility for AI-Related Patent Applications

Last week, the USPTO published a rare Appeals Review Panel (ARP) decision in Ex parte Desjardins, Appeal 2024-000567, September 26, 2025 (“ARP Decision”), that reversed a finding of ineligible subject matter in a patent application directed to training of machine learning models. This decision, authored by new USPTO Director John Squires in the first week of his tenure, appears to signal new USPTO leadership's intent to limit examiners' reliance on subject-matter eligibility to reject claims, particularly when read in combination with the memo of August 4, 2025, from Deputy Commissioner for Patents, Charles Kim. 

The patent application at issue (U.S. Patent Application No. 16/319,040) was directed to training of a machine learning model to perform multiple different tasks. A representative claim read:

1. A computer-implemented method of training a machine learning model,
wherein the machine learning model has at least a plurality of parameters and has
been trained on a first machine learning task using first training data to determine first values of the plurality of parameters of the machine learning model, and
wherein the method comprises:
determining, for each of the plurality of parameters, a respective measure of an
importance of the parameter to the first machine learning task, comprising:
computing, based on the first values of the plurality of parameters
determined by training the machine learning model on the first machine learning
task, an approximation of a posterior distribution over possible values of the
plurality of parameters,
assigning, using the approximation, a value to each of the plurality of parameters, the value being the respective measure of the importance of the parameter to the first machine learning task and approximating a probability that the first value of the parameter after the training on the first machine learning task is a correct value of the parameter given the first training data used to train the machine learning model on the first machine learning task;
obtaining second training data for training the machine learning model on a second,
different machine learning task; and
training the machine learning model on the second machine learning task by training the machine learning model on the second training data to adjust the first values of the plurality of parameters to optimize performance of the machine learning model on the second machine learning task while protecting performance of the machine learning model on the first machine learning task,
wherein adjusting the first values of the plurality of parameters comprises adjusting the first values of the plurality of parameters to optimize an objective function that depends in part on a penalty term that is based on the determined measures of importance of the plurality of parameters to the first machine learning task.

The patent examiner rejected the claims as being obvious under 35 U.S.C. §103 but did not reject the claim as being directed to patent-ineligible subject matter under 35 U.S.C. §101. On appeal, the Patent Trial and Appeal Board (PTAB) affirmed the rejection under §103 and introduced a new ground of rejection under §101. Director Squires exercised his authority to convene an ARP to reconsider the PTAB Decision.   

The ARP left intact the §103 rejection but vacated the §101 rejection introduced by the PTAB. Applying the two-step patent eligibility inquiry as set forth in Alice Corporation v. CLS Bank International, 573 U.S. 208 (2014), and in the Manual for Patent Examining Procedure (MPEP) at §2106, the ARP agreed with the PTAB that each claim recited a judicial exception, specifically, “at least one abstract idea.” (ARP Decision at 6-7).   

Moving to the next prong of the inquiry, the ARP found that the PTAB had erred. The claims did “integrate the judicial exception into a practical application” as provided by MPEP §2106.04(II)(A)(2). The ARP relied on Enfish, LLC v. Microsoft Corp., 822 F.3d 1327, 1339 (Fed. Cir. 2016), in which “the Federal Circuit held that the eligibility determination should turn on whether ‘the claims are directed to an improvement to computer functionality versus being directed to an abstract idea.'” ARP Decision at 8, citing Enfish, 822 F.3d at 1336. The claim language reciting “adjust the first values of the plurality of parameters to optimize performance of the machine learning model on the second machine learning task while protecting performance of the machine learning model on the first machine learning task” was found to “constitute[] an improvement to how the machine learning model itself operates, and not, for example, the identified mathematical calculation.” ARP Decision at 9.

Notably, the ARP Decision also included language directing examiners and PTAB panels on how to evaluate patent eligibility under §101:

Under a charitable view, the overbroad reasoning of the original panel below is perhaps understandable given the confusing nature of existing § 101 jurisprudence, but troubling, because this case highlights what is at stake. Categorically excluding AI innovations from patent protection in the United States jeopardizes America's leadership in this critical emerging technology. Yet, under the panel's reasoning, many AI innovations are potentially unpatentable-even if they are adequately described and nonobvious-because the panel essentially equated any machine learning with an unpatentable "algorithm" and the remaining additional elements as "generic computer components," without adequate explanation. Dec. 24. Examiners and panels should not evaluate claims at such a high level of generality.

…

At the same time, the claims at issue stand rejected under § 103. This case demonstrates that §§ 102, 103 and 112 are the traditional and appropriate tools to limit patent protection to its proper scope. These statutory provisions should be the focus of examination.

ARP Decision at 9-10.

With the publication of this Decision, the new USPTO Director appears to be signaling a policy shift. In particular, improvements in the operation of machine learning models (and potentially other complex computer algorithms) should be considered as patent-eligible improvements to the functionality of a computer system. The more general statement that §§ 102, 103, and 112 (novelty, non-obviousness, and sufficiency and clarity of the disclosure and claims) should be the focus of examination may carry over into other technologies. 

While this ARP Decision may result in the USPTO finding more claims eligible under §101, applicants should keep in mind that the USPTO's decisions are not binding on the courts. The Federal Circuit most recently reiterated this point in Rideshare Displays, Inc. v. Lyft, Inc., No. 2023-2033 (Fed. Cir. Sept. 29, 2025) (nonprecedential).

Takeaways for evaluating inventions for patenting:

1. Claims that recite an improvement in the operation of a machine-learning model are now more likely to be considered patent-eligible by the USPTO.

2. Patent examiners may begin to focus more on questions of novelty and non-obviousness and less on subject-matter eligibility. Applicants may receive fewer rejections under 35 U.S.C. §101 or find them easier to overcome. This change may extend beyond machine-learning or computer-implemented inventions.

3. Courts are not bound by USPTO interpretations of statutes. While it may become easier to obtain a patent in some technology areas, enforcement will likely remain a challenge.