This study analyzes the impact and practical implications of the USPTO’s 2024 AI Artificial Intelligence Subject Matter Eligibility Guidance (AI SME). It was announced on July 17, 2024, on the patent eligibility criteria for AI-related inventions, in comparison with the 2019 Patent Eligibility Guidance (PEG). The 2024 AI SME builds on the analytical framework of the 2019 PEGwhile imposing stricter eligibility criteria for AI-related inventions. In Step 2A, Prong 1, such inventions are more likely to be classified as mathematical concepts or mental processes, thereby increasing the likelihood of being deemedabstract ideas. Consequently, overcoming rejections under Step 2A, Prong 1, becomes more difficult. Furthermore, the 2024 AI SME provide no examples, additional explanations, or case-law updates demonstrating how AI-related inventions can satisfy Step 2B, suggesting its diminished significance in the analysis. Consequently, for AI-related inventions, the Alice/Mayo framework effectively condenses to Step 2A, Prong 2, where demonstrating technical improvementis the central issue. Proving technical improvement requires claims and specifications to clearly articulate technical specificity, often encouraging narrower claim scopes. Accordingly, the 2024 AI SME apply more stringent examination standards to AI-related inventions than to others. While the 2024 guidance enhances clarity by including numerous examples and analyses, unresolved issues persist, such as conflicts between Example 39 of the 2019 PEG and Example 47 of the 2024 AI SME. Further, the lack of specific examples illustrating technical improvements in AI technology limits the practical effectiveness of the standards.

However, the Memorandum (Reminders on Evaluating Subject Matter Eligibility of Claims under 35 U.S.C. § 101, 2025 Memo), issued by the USPTO on August 4, 2025, partially relaxes the stringent standards established by the 2024 AI SME. The 2025 Memo offers nuanced interpretations of mathematical relationships, mental processes, technological advancements, the preponderance of evidence standard, and also provides additional clarification for Examples 39 and 47. In doing so, it presents practitioners with critical guidance for addressing patent eligibility issues to AI-related inventions.