Over the past few months, the U.S. Patent and Trademark Office has been treating artificial intelligence and machine learning inventions more favorably under 35 U.S.C. 101 than in years.
This change did not occur overnight, but took shape through a series of developments around the incident. Expert Desjardinsculminating in new guidance that reshapes the way examiners and claimants alike think about the eligibility of AI-related claims.
turning point
In September 2025, the PTO caused a stir when the Appellate Review Board, under Director John Squires, reversed a Section 101 denial. desjardins. Applications related to machine learning had been rejected simply because their claims were abstract.
But Squires agreed with inventor Guillaume Desjardins that the technology improves “the way machine learning models themselves operate, rather than, for example, specific mathematical calculations.”
Members of the Appellate Review Board agreed, saying the claims describe meaningful technical improvements to the way machine learning models operate, including reducing system complexity, improving multitasking learning, and preventing “catastrophic forgetting.” The Commission further explained that such improvements may meet the “practical application” requirement under the Act. alice Patent eligibility test and under § 101.
The message was clear from the beginning. AI innovation should not be reflexively categorized as an abstract idea. The committee warned that categorically excluding AI innovation from patent protection risks undermining U.S. leadership in important emerging technologies.
rising desjardins
The PTO has taken the following additional actions: Expert Desjardins This wasn’t just symbolic; it was what it meant. desjardins It will now become the binding authority across the PTO, formally guiding both review and appeal decisions.
He also emphasized points that experts have been advocating for years. In other words, claims that improve the functionality of a computer or another technology, even through logical structures or processes rather than physical hardware, may be patent eligible under Section 101.
Such reasoning is based on longstanding authority in the United States Court of Appeals for the Federal Circuit. Enfish LLC v. Microsoft Corp.. A decision that recognizes that software-based improvements may not be abstract.
by increasing desjardinsthe PTO has indicated that it wants examiners to evaluate Section 101 eligibility through the lens of technical contribution rather than recursive abstraction. Claims that describe specific architectural, performance, or complexity improvements to AI and machine learning systems are now firmly within the scope of patent eligibility.
systematic change
The USPTO took the next step in December 2025, issuing a formal memorandum that consolidated: desjardins It will be incorporated into the Manual of Patent Examination Procedures (MPEP). This update is significant because it changes how Section 101 is taught and applied in practice.
The revised MPEP directs examiners to evaluate whether a claim refers to an improvement in computer functionality or other technical field when assessing eligibility under Step 2A. alice framework.
Examiners should look for specific advantages set forth in the claims and supported by the specification, rather than issuing rejections of abstract ideas that are too general or too broad. Examples include increasing model training efficiency, reducing complexity, improving data structures, and other technological advances.
This memo makes clear that qualifications must be consistent with existing guidance. Traditional patentability principles such as novelty (Article 102), nonobviousness (Article 103), and written description (Article 112) remain the primary mechanisms for regulating claim scope. Eligibility does not specifically exclude software or AI, but is framed as a technical inquiry into how the invention works.
What this means
Taken together, these three developments represent a significant change in the way the PTO handles subject matter eligibility for AI and machine learning inventions.
The lesson for practitioners and applicants is that successful AI patenting is increasingly about advocating for and supporting specific technological improvements, rather than just high-level outcomes or the use of AI in a business context.
Claims that describe how a model is trained, how its architecture improves performance, or how processing is optimized are now more likely to survive Section 101 scrutiny.
This trend is already visible in subsequent decisions by the Patent Trial and Appeal Board. in carmody sidethe PTAB reversed the Section 101 denial on December 30. Because the claims described a modular machine learning architecture with defined training datasets and plug-and-play models that improve not only what the system accomplished, but how the system operates.
That reasoning closely follows desjardins and highlights new directions for Section 101 analysis regarding AI technologies.
Current situation
meanwhile desjardins This does not completely eliminate Section 101 scrutiny, but rather reframes it in a more technology-focused manner that is consistent with both Federal Circuit precedent and the PTO’s modernization goals. For AI innovators, this means that careful claim drafting and robust technical disclosures can meaningfully shift eligibility analysis to their advantage.
in short, desjardins It has evolved from a single appellate intervention to the foundation of modern Section 101 practice regarding AI technology. Its effects are already shaping the PTAB outcome and may continue to do so in 2026 and beyond.
This article does not necessarily reflect the opinion of Bloomberg Law, Bloomberg Tax, Bloomberg Government, the publisher Bloomberg Industry Group, Inc., or its owners.
Author information
Matthew R. Carey is a partner and chair of Marshall, Gerstein & Borun’s electrical and computer technology practice group.
Please write to us: Author guidelines
