IP Practice Group Leader, Jeff Andrews, was prominently featured in a recently published article from BioWorld MedTech that analyzed the USPTO Director’s Appeals Review Panel (ARP) decision in ex parte Desjardins, a notable win for machine learning subject matter eligibility.
Recently, Google sought patent protection for a machine learning invention addressing catastrophic forgetting in AI systems. Their application was rejected as obvious, but on appeal, the Patent Trial and Appeal Board (PTAB) added a Section 101 rejection for being too abstract. The ARP opinion was strongly critical of the PTAB’s Section 101 analysis and vacated the subject matter ineligibility finding, looking to the specification’s description of how the claimed invention would improve the functioning of an AI model.
Andrews explained why Director John Squires’ intervention was no accident, tying it to the PTO’s August 2025 internal guidance emphasizing the proper two-step Section 101 analysis and to longstanding Federal Circuit precedent like Enfish v. Microsoft.
He underscored the decision’s broader implications beyond software, potentially reaching diagnostics, and frames it as a policy recalibration, instructing the PTO and PTAB to stop conflating ML with abstract algorithms.
Andrews distilled the practical takeaway for applicants: substance over labels.
Eligibility turns on integrating abstract ideas into a practical application that improves computer technology or another technical field—so AI/ML filers must “do the work” in the specification and claims by explaining concrete, system-level benefits rather than functional aspirations.
While noting that the Federal Circuit will continue to police the boundary, Andrews highlighted how Desjardins provides near-term clarity inside the PTO and signaled renewed seriousness about rigorous, technology-grounded Section 101 analysis.
Full coverage is available at BioWorld MedTech (paywall).