USPTO Implements Sweeping Changes to Software Invention Patent-Eligibility under 35 U.S.C. 101

For just over a decade, patent applicants – especially those focused on developments in the software space – have had their inventions subjected to heavy scrutiny under 35 U.S.C. § 101 due to the vague multi-step framework set in place by the Supreme Court’s decision in Alice Corp. v. CLS Bank International. While the outcome of this case failed to define terms pertinent to patent-eligibility analysis (such as “abstract idea”), it succeeded in creating an abnormally high hurdle for inventors and patent practitioners alike, especially when the focus of an application’s claims were directed towards software, artificial intelligence, and the use of “generic” computer systems.

This multi-step system required Examiners at the United States Patent and Trademark Office (USPTO) to determine whether the claims of a patent application were directed to an unpatentable law of nature, natural phenomena, or abstract idea. If the claims fell into any of these categories, the second step was to determine whether the claims nonetheless included an inventive concept that is sufficient to “transform the nature of the claim into a patent-eligible application.” I took a much deeper dive into the standards for patent-eligibility and their relationship to software innovations in my previous blog, for those that wish to engross themselves into the particulars on how software inventions were subjugated to the lengthy, yet ill-defined process for determining whether patent-eligibility was present within the respective claims. The Manual of Patent Examining Procedure provides a flowchart to help visualize this multi-step process in Section 2106, which is shown below:

Last summer, the USPTO, in what some practitioners (myself included) deemed to be a surprising move, issued a memorandum on August 4, 2025 with the explicit disclaimer that “[t]his memorandum is not intended to announce any new USPTO practice or procedure and is meant to be consistent with existing USPTO guidance.” Upon a closer read of this memorandum however, I couldn’t help but find myself wondering whether the USPTO, facing at that time what many would agree was a new technological revolution fueled by the rapid development of artificial intelligence, was encouraging their Examiners to – for lack of a better cliché – “shift gears” when examining patent applications in the software space.

In summary, that memorandum recognized that, when a software or AI claim purported to improve the capabilities of a computer or other technology, it was reciting a practical application, and therefore should not automatically be deemed patent-ineligible simply because it was involved with generic computer components.

The memorandum reminded examiners that “if it is a ‘close call’ as to whether a claim was eligible, they should only make a rejection when it is more likely than not (i.e., more than 50%) that the claim is ineligible under 35 U.S.C. § 101. “A rejection of a claim should not be made simply because an examiner is uncertain as to the claim’s eligibility…unpatentability must be established by a preponderance of the evidence.” The USPTO’s sentiment regarding examiners’ prior application of patent-eligibility rejections in software related applications appeared consistent throughout this memorandum – they were too frequent/loosely applied, and therefore needed to be dialed back. At the time, I reminded myself that this memorandum was meant to be nothing more than “consistent with existing USPTO guidance” as so stated, but I couldn’t help but feel a sense of cautious optimism that software-based patent applications would begin to see more of a fair shake during patent prosecution and appeal proceedings when compared to their non-software related counterparts.

Last September, that cautious optimism started to shape itself into confirmation with the issuance of the In re Desjardins decision. The USPTO Appeals Review Panel – formed by the newly elected Director, the Commissioner for Patents, and the Chief Administrative Patent Judge – reviewed and vacated a decision by the Patent Trial and Appeal Board that upheld a patent-eligibility rejection against claims directed to an AI model trained on new tasks in succession while protecting knowledge about prior tasks. I’ll leave the specifics of this claim language and how the AI model was trained out of this blog in an effort to keep things on the “lighter” side, but if you’re curious, I’ll provide the link to the Board’s discussion and citation to the exemplary independent claim in the below footnotes.

The Appeal Review Panel didn’t just reverse the PTAB’s overextending application of the 101 patent-eligibility rules, however. They also presented a solid stance on why the prior application of 101 was in desperate need of an update (poor pun intended). First, they noted that claim limitations which recited improvements to computer technology or to the technological field (in this case, machine learning) under a Step 2A Prong 2 analysis, would push the claims into the realm of patent-eligibility. Furthermore, they provided what I believe to be the most important statement of their entire decision: “categorically excluding AI innovations from patent protection in the United States jeopardizes America’s leadership in this critical emerging technology.”

The Desjardins decision not only reminded patent practitioners and examiners alike that improvements to computer technology included improvements to software, but also highlighted the recognition that software technology is stepping into a new era, and the innovation spurring this AI revolution can no longer be ignored as an unpatentable concept.

In case the Desjardins decision wasn’t convincing enough to practitioners and inventors that the USPTO was finally making a shift, a December 4th Memo issued by the Director further confirmed that the Desjardins decision would now be precedent, to “ensure the case reasoning binds all examination and appeals activity”, and to “underscore that improvements in computational performance, learning, storage, data sets and structures, for example, can constitute patent-eligible technological advancements under the Alice framework.” The December 4th Memo further provided suggestions for practitioners to help advance their software applications to allowability, including the encouraged use of inventor declarations as an option to supply further evidence that software related applications do in fact present an improvement to the technological field they pertained to.

In summary, we are entering a new age of examination at the Patent Office, where software-based patent applications may finally get their equal chance at achieving allowance. I implore those inventors that are “on the fence” with whether to file for a patent on their software innovations to now strongly consider doing so in view of this rule shift. The window for opportunity may not stay open forever, as I can only surmise that the federal courts will have their turn to review and opine on this rule shift within the next several years.