The Appeals Review Panel (ARP), led by USPTO Director John A. Squires, ruled in Ex parte Desjardins that patent requests for technical improvements in Artificial Intelligence (AI) technology meet the basic eligibility requirement under 35 U.S.C. § 101. This precedential ruling means AI inventions cannot be automatically rejected as abstract ideas. The ARP set aside a new § 101 rejection that the Patent Trial and Appeal Board (PTAB) had created, which targeted patent claims about training machine learning models.
What Did the Appeals Review Panel Overturn in This Case?
The ARP decision led to two main results:
- Reversal of the Rejection: The ARP overturned the PTAB’s finding that claims directed to improving machine-learning models were ineligible under § 101, determining instead that the claims meet the patent eligibility requirement.
- Nonobviousness Still Stands: The ARP left in place the original rejection of the claims under § 103. The claims must still be proven truly inventive to be granted, confirming § 101 is not the final hurdle.
The ARP emphasized that the claims were eligible because they solved a technical problem, stating that “categorically excluding AI innovations from patent protection in the United States jeopardizes America’s leadership in this critical emerging technology.”
How Must AI Claims Be Written to Satisfy Eligibility?
The ARP reaffirmed the importance of the Federal Circuit’s Enfish precedent. For a claim to be eligible, it must focus on improving the computer or technology itself, rather than merely stating an abstract idea.
The claims in Ex parte Desjardins satisfied this standard because they focused on a specific technical solution:
- The Improvement: The claimed method is to “effectively learn new tasks in succession whilst protecting knowledge about previous tasks” and enable “reduced system complexity.”
- The Enfish Link: This technical result demonstrated that the invention was defined by “logical structures and processes,” which the ARP found sufficient to conclude the claims were not directed to an abstract idea.
What are the Main Rules for Granting Patents?
The ARP decision emphasizes that the § 101 rule for eligibility is only meant to filter out purely abstract concepts. The decision confirms that the following rules are the correct and primary focus for patent examination:
| Statutory Requirement | Legal Focus |
| Novelty (35 U.S.C. § 102) | Determines if the invention is new compared to prior art. |
| Nonobviousness (35 U.S.C. § 103) | Determines if the invention is sufficiently inventive beyond what is obvious to an expert. |
| Disclosure (35 U.S.C. § 112) | Ensures the patent clearly describes the invention and how to use it. |
The ARP stated that §§ 102, 103, and 112 are the “traditional and appropriate tools to limit patent protection to its proper scope,” and these statutes “should be the focus of examination.”
What is the Impact of This New Rule?
This decision tightens § 101 practice at the USPTO and discourages examiners from using it as a catchall to block otherwise patentable claims. This is a crucial development for those in the computer and AI space who have been facing inconsistent § 101 rejections.
- Certainty for Innovators: Companies and inventors now have a clearer, more reliable standard for demonstrating the patent eligibility of their AI innovations.
- For Legal Practitioners: Provides patent practitioners with a direct and binding administrative tool to challenge § 101 rejections for AI and software claims by focusing arguments on demonstrable technical improvements under the Enfish standard.
- Applicant Strategy: Patent applications must now strongly emphasize the technical advantage the invention provides to the machine learning model or computer system.
This development clarifies the application of patent eligibility standards, ensuring the examination process is focused on the substantive requirements of novelty and nonobviousness.
