USPTO Issues New Guidance on Patent Subject Matter Eligibility under 35 U.S.C. § 101
On August 4, 2025, the USPTO issued critical guidance that could significantly reduce § 101 rejections for AI and software patents. This memorandum[1] reinforces existing law but provides examiners with clearer boundaries and a higher standard for making eligibility rejections. For companies developing AI, machine learning, and software technologies, this guidance represents a more favorable examination environment and opportunities to overcome previously problematic rejections.
In addition to requirements of novelty, nonobviousness, and sufficient written description, patents must be directed to patent eligible subject matter. That is, regardless of how revolutionary the invention may be, if it is directed to a law of nature, natural phenomena, or abstract idea, collectively referred to as “judicial exceptions,” the patent may be rejected[2]. The abstract idea category is ill-defined, but has been described as including mathematical concepts[3], certain methods of organizing human activity[4], and mental processes[5]. This last category can be particularly problematic for software-related patents, as some patent examiners argue that program code can be performed laboriously with pen and paper, and therefore are merely mental processes regardless of complexity. This has led to a surge in eligibility rejections as high as 80% in some art units, particularly those dealing with inventions related to artificial intelligence (AI). The inconsistent application of eligibility standards has created uncertainty in the patent landscape, with identical or similar AI technologies receiving different treatment depending on the examiner assigned. This unpredictability has made it difficult for companies to develop coherent patent strategies and accurately budget for prosecution costs.
The “Two Prong” Test for Whether Claims Are Directed to a Judicial Exception
Patent examiners determine whether a claim is directed to a judicial exception by applying a two prong test (sometimes referred to as Step 2A prong one and Step 2A prong two of the Alice/Mayo test). Under prong one, the examiner determines whether a claim sets forth or describes an abstract idea, law of nature, or natural phenomenon. If so, under prong two, the examiner determines whether the claim as a whole integrates the recited judicial exception into a practical application of the exception. If it does, the claim is eligible under § 101 (and may be patentable, subject to the remaining requirements).
This two prong test is not always applied consistently or carefully, and examiners have at times treated all software claims as mental processes, and focus only on hardware elements such as a processor or memory recited in the claim to argue that there is no integration of the mental process into a practical application. The new memo addresses both of these issues.
First, the memo reminds examiners not to expand the mental process category to read on claim limitations that cannot practically be performed in the human mind, and states that “claim limitations that encompass AI in a way that cannot be practically performed in the human mind do not fall within this grouping.”[6]
Second, the memo reiterates that the prong two analysis “considers the claim as a whole… Accordingly, the additional limitations should not be evaluated in a vacuum, completely separate from the recited judicial exception. Instead, the analysis should take into consideration all the claim limitations and how these limitations interact and impact each other when evaluating whether the exception is integrated into a practical application.”[7]
Importantly, the memo establishes a concrete threshold for making eligibility rejections. Examiners must now demonstrate that it is "more likely than not"—greater than 50% probability—that a claim is ineligible[8]. This represents a meaningful departure from previous practice where examiners could reject claims based on uncertainty alone. The preponderance of evidence standard creates a presumption in favor of eligibility for borderline cases, potentially reducing frivolous rejections by thousands annually.
Critical Distinction: “Recites” Versus “Involves” a Judicial Exception
The memo emphasizes a crucial distinction that can determine patent eligibility: whether claims “recite” a judicial exception versus those that merely “involve” one[9]. Claims that only involve mathematical concepts or abstract ideas—without explicitly setting forth the mathematical relationships or algorithms—remain eligible and do not require further analysis. For example, a limitation stating “training a neural network using a training dataset” involves mathematical concepts but does not recite them, making the claim more likely to be eligible. In contrast, limitations that specifically name algorithms like “wherein the training algorithm includes a backpropagation algorithm” or “a gradient descent algorithm” recite mathematical concepts and require additional analysis for eligibility.
Practical Implications: What Examiners Should No Longer Reject
The memo provides concrete guidance on AI limitations that should not be categorized as mental processes. Complex neural network operations, multidimensional matrix calculations inherent in deep learning models, and hardware-specific AI implementations that require specialized processors cannot practically be performed by humans and should not face mental process rejections. This clarification is particularly significant for patents involving large language models, computer vision systems, and machine learning hardware accelerators.
What This Means for Patent Applicants
The new memo does not represent a change in the law, but it may signal a welcome shift in examination procedure that will be favorable to companies innovating in software, AI, and machine learning. Eligibility rejections, which are particularly frustrating when an examiner confirms that no relevant prior art would otherwise bar patentability, may be rarer in the future. With examiners being told to practice restraint and avoid applying broad eligibility rejections, software patent applicants may find the examination process smoother and more successful.
Both in addressing rejections and in drafting patent applications from the outset, applicants should consider emphasizing how claim limitations cannot be practically performed in the human mind; how the claims provide specific technological improvements over the prior art; and how the claims integrate any judicial exceptions into concrete, technical applications. We recommend working closely with counsel to craft claim language and prosecution strategies that highlight technological improvements and practical applications.
Strategic Opportunities for Patent Applicants
This guidance creates immediate opportunities for both pending applications and future filings:
For Pending Applications: Applications currently under examination should be reviewed for potential arguments based on this guidance. Responses to § 101 rejections can now leverage the "close call" standard, argue that AI-specific limitations cannot be performed mentally, and emphasize holistic claim analysis. RCE filings and appeals may benefit from citing this memo's specific requirements for examiner analysis.
For New Filings: Claim drafting strategies should emphasize technical implementations that clearly cannot be performed mentally, avoid unnecessary recitation of named algorithms unless essential, and include robust dependent claims with specific hardware or software improvements. Focus on particular solutions rather than general outcomes, and ensure specifications describe technical improvements that would be apparent to skilled artisans.
For Portfolio Management: Companies should reassess previously abandoned applications that received § 101 rejections, particularly those involving AI or machine learning technologies. Continuation applications may now have better success rates under this guidance.
For questions regarding how the USPTO’s memorandum may affect your pending applications or prosecution strategy, please contact partners Daniel Rose, Intellectual Property Department, or Sid Kapoor, Intellectual Property Department.
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] “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).
[2] Diamond v. Chakrabarty, 447 U.S. 303, 309 (1980).
[3] Gottschalk v. Benson, 409 U.S. 63, 71-72 (1972).
[4] 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).”).
[5] 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)).
[6] “Reminders on evaluating subject matter eligibility of claims under 35 U.S.C. 101” at 2.
[7] Id. at 3.
[8] Id. at 5.
[9] Id. at 3.
