USPTO’s Subject Matter Eligibility Declarations Reshape AI Patent Section 101 Review

On December 4, 2025, the United States Patent and Trademark Office (USPTO) issued two memoranda from Director John A. Squires introducing “Subject Matter Eligibility Declarations (SMEDs),” a structured framework comprising best practices guidance for applicants and examination guidelines for patent examiners. Concurrently, the USPTO abolished most of the previous administration’s AI patent guidance and established a new framework stating that “no separate subject matter eligibility standards apply” to AI-identified inventions. These changes fundamentally reshape patent prosecution practices under 35 U.S.C. § 101 for algorithm-related inventions and machine learning technologies.

Structure of SMEDs and Applicant Options

SMEDs provide patent applicants with a structured evidence framework to counter subject matter eligibility rejections. Applicants may now submit evidence categorized into four distinct types.

Four Categories of Admissible Evidence

The first category encompasses evidence demonstrating that steps recited in a claim cannot be performed by the human mind alone. This evidence establishes that the invention is inherently dependent on computer processing.

The second category includes evidence showing that the invention substantially improves computer function or a related technological field. This category addresses innovations directed to specific technical domains rather than generalized computational processing.

The third category pertains to evidence demonstrating that medical preventative and therapeutic steps integrate abstract concepts into practical application. Biomedical inventions, particularly diagnostic and therapeutic methods, are now better positioned to establish eligibility through this evidence category.

The fourth category involves evidence that an unconventional arrangement of conventional elements achieves a significant technical improvement beyond what constitutes judicial exceptions. System configurations and process flow innovations are paramount under this standard.

Restructuring the Alice/Mayo Framework in Practice

Eligibility determinations in U.S. patent examination have long relied on the two-step test established in Alice v. CLS Bank (2014) and Mayo v. Prometheus (2012). Step one evaluates whether a claim falls within judicial exceptions such as abstract ideas, laws of nature, or natural phenomena. If it does, step two assesses whether additional elements, “considered as a whole,” are sufficient to transform the exception into a patent-eligible invention.

SMEDs enable applicants to systematically submit objective evidence, expert declarations, and experimental results at each stage of this two-step analysis. What was previously subject to examiner discretion and subjective judgment is now anchored to documentary evidence and technical foundation. This represents a significant procedural modernization in patent examination practice.

Transformation of AI Patent Eligibility Policy

The USPTO’s elimination of most prior AI patent guidance on November 28, 2025, provides essential context for understanding SMEDs’ introduction. The new framework explicitly provides that AI-identified inventions are not subject to “separate subject matter eligibility standards.” This means AI-implemented inventions are evaluated under the same framework as conventional software inventions.

However, this does not necessarily ease eligibility analysis for AI patents. Rather, it establishes a unified and consistent examination standard requiring objective evidence-based evaluation. SMEDs serve as the practical implementation mechanism for this framework transition, providing structure and consistency to how such evidence is assessed.

Ex parte Desjardins: A Precedential Harbinger

On September 26, 2025, the Patent Trial and Appeal Board’s Appellate Panel Reconstituted (APR) issued a significant decision in “Ex parte Desjardins.” Director John A. Squires participated in this decision, which reversed a Section 101 rejection of a machine learning-related patent that had been previously rejected. This precedential decision signaled the policy direction that SMEDs would subsequently formalize, providing industry guidance before the formal announcement.

Practical Impact on Patent Applicants

SMEDs introduction carries substantial implications for applicant prosecution strategy across multiple contexts.

First, early-stage Office Action responses are strengthened. Upon receiving a rejection, applicants can now craft substantive rebuttals grounded in the structured evidence framework, enabling more productive dialogue with examiners.

Second, international filing strategy optimization is facilitated, particularly for U.S.-Japan synchronized prosecution. Japanese applicants can now leverage experimental data and technical disclosures submitted in Japanese applications as SMED evidence through Rule 132 declarations in U.S. applications. This integration enhances global filing efficiency.

Third, relative patent eligibility improvements for AI and machine learning inventions become achievable. Examination of AI inventions, previously discretionary and unpredictable, transitions toward evidence-based evaluation, reducing uncertainty in patent prosecution outcomes.

Design Considerations and Limitations

SMEDs’ efficacy depends fundamentally on applicants’ ability to prepare and submit high-quality evidence. Generic technical descriptions and sweeping generalizations remain unpersuasive under the new framework. Instead, concrete experimental results, expert opinions from relevant technical fields, and quantifiable technical benefits must be demonstrated substantively.

Furthermore, Thaler v. Vidal precedent mandates that artificial intelligence systems cannot be designated as inventors under U.S. patent law. Even for AI-related inventions, human inventor identification and demonstrable creative contribution remain legally essential. This requirement persists unchanged post-SMEDs.

Future Development and Industry Implications

SMEDs represent a structured procedural response to the long-standing problem of uncertainty in software and AI patent eligibility determinations. However, they do not eliminate subjectivity entirely; rather, they place increased responsibility on applicants to prepare and present substantive evidence. The burden of proof framework shifts toward evidence-intensive prosecution.

Leading intellectual property firms, including Morrison & Foerster and Venable LLP, have emphasized the centrality of SMED strategy for 2026 patent prosecution, with industry-wide protocol development underway.

Sources

• USPTO Official Announcement: “USPTO Issues New Guidance on Subject Matter Eligibility Declarations”
• IP Watchdog: “USPTO Solves U.S. Patent Eligibility Problems”
• Morrison & Foerster: “USPTO Puts Subject Matter Eligibility Declarations in the Spotlight”
• Venable LLP: “The 101 Reset for 2026: New USPTO Guidance on AI Eligibility and When Early Motions Matter”