Legal industry analyst Nicola Shaver identified Landfall IP as one of the leading AI-native law firms globally—and the only patent practice on the list.
November 19, 2025
BY Nathan Mutter
Considering filing a patent on AI-related inventions?
Here is the TL;DR on recent USPTO developments and what you need to know:
Read below for more details and insight:
One week into his tenure as USPTO Director, John Squires personally intervened to reverse a § 101 (subject matter eligibility) rejection of an AI patent application and issued a pointed rebuke to his own Patent Trial and Appeal Board (PTAB) in the process.
This remarkable action points to a deliberate policy signal that AI patents will no longer face the categorical skepticism they’ve endured for years.
While this shift is favorable for innovators pursuing AI-related patents, both patentees and their counsel must carefully navigate the complex, rapid evolution of the interplay between this new USPTO policy and existing federal court precedent.
Global Competition and Shifting Policies
Director Squires has been explicit about his goals. In testimony before the Senate Subcommittee on Intellectual Property, he stated that § 101 “should not be misused as a blunt instrument to exclude entire technological fields,” warning that doing so “risks disqualifying exactly the kinds of advances America needs most.”
His motivation extends beyond abstract fairness. Squires emphasized that “without reliable patent protection, AI start-ups cannot secure the venture capital needed to compete against state-backed giants in China and elsewhere.” For him, patent eligibility is directly tied to maintaining U.S. leadership in AI innovation.
The USPTO has backed this rhetoric with action. An August 2025 memo reminded examiners of all the ways to overcome § 101 rejections for AI and software inventions and cautioned against oversimplifying claim limitations. PTAB data from October 2025 shows a dramatic reduction in new § 101 grounds of rejection, with reversal rates of examiner rejections at unprecedented levels.
Squires’ directive is simple: use § 102 (novelty), § 103 (obviousness), and § 112 (enablement) as the appropriate gatekeepers—not § 101.
Key Points from Ex parte Desjardins (the case Squires reversed)
In March 2025, the PTAB entered a new ground of rejection under 35 U.S.C. § 101 against DeepMind Technologies’ patent application directed to methods for training machine learning models on multiple sequential tasks while avoiding a fundamental ML challenge referred to as “catastrophic forgetting.”
The PTAB found the claims were directed to an abstract “mathematical algorithm” and therefore ineligible for patent protection.
Director Squires vacated the PTAB’s decision using the rarely-deployed Appeals Review Panel. The decision carried unusually pointed criticism, stating that the PTAB “eschewed the clear teachings of Enfish” and “substituted only a cursory analysis that ignored this well-settled precedent.”
More significantly, the decision warned that “categorically excluding AI innovations from patent protection in the United States jeopardizes America’s leadership in this critical emerging technology.”
Squires seems to be sending a clear message: the USPTO’s approach to AI patents is changing.
But Federal Courts Still Matter
Before AI founders celebrate too enthusiastically, it’s worth noting an important caveat: while the USPTO may grant the patent, there remains a possibility that a court could later invalidate it.
Federal courts continue to apply rigorous scrutiny to patent eligibility. In a recent 2025 decision, Judge William Bryson of the Federal Circuit (sitting by designation in district court) synthesized § 101 jurisprudence into a six-part framework that invalidated claims to improving online dating profiles through AI analysis.
Judge Bryson’s framework highlights the key distinction that courts enforce: claims directed to actual improvements in computer technology can be patent-eligible, while claims that merely apply AI to existing processes using generic computing resources typically are not.
This isn’t unique to Bryson’s decision—it reflects the stance of many in the judiciary. The Recentive Analytics case from April 2025, where the Federal Circuit rejected machine learning patents as “doing it on AI,” demonstrates that courts remain skeptical of claims that don’t articulate specific technical improvements.
The result is a divergence: the USPTO is becoming meaningfully more permissive, while federal courts maintain strict eligibility standards. AI companies and their patent attorneys must navigate both.
The Path Forward: Technical Improvements Are Key
So what does this mean in practice?
The Desjardins claims survived because they addressed a specific technical problem—catastrophic forgetting—with a technical solution. The claims focused on how the ML system operates, not just what it accomplishes.
This is the through-line that connects USPTO practice and court precedent: frame your innovations as technical improvements to AI methodology, not just novel applications of existing techniques.
For AI innovations, it’s important to emphasize:
What This Means for AI Founders
For AI startups, this represents a genuine opportunity.
The good news: USPTO examination of AI innovations is expected to become more predictable and less hostile, reducing the time and cost of overcoming § 101 rejections. Examiners are now encouraged to focus their review on prior art and enablement, rather than abstract eligibility.
The practical reality: Despite recent USPTO policy changes, a strong patent strategy remains vital because federal courts, which are not bound by these shifts, will rigorously scrutinize any litigated patent. Therefore, your claims must satisfy both the newly permissive USPTO standards and the stringent requirements of federal courts.
If you’re building AI infrastructure, novel training methods, or architectural innovations, now is the time to file. The USPTO window of receptiveness is open, and the competitive landscape for AI patents is still forming. Companies that establish strong, well-drafted portfolios now will have significant leverage as the space matures.
The § 101 landscape for AI patents has fundamentally shifted. Are you ready to take advantage of it?
Our Discovery Agent will work with you to explore whether your invention has what it takes, and you’ll get a detailed report to help you with the next steps. Request a demo to experience this transformational tool.
Discovery Agent
