In a precedential decision issued by the U.S. Court of Appeals for the Federal Circuit in February 2025, the court affirmed a district court ruling that the term “payment handler” in a patent claim was a “nonce” term—a placeholder for functional language. This ruling invoked 35 U.S.C. § 112, sixth paragraph, governing means-plus-function claiming, leading to the invalidation of the associated patents. The decision raises crucial questions about the drafting of patent claims in industries like artificial intelligence (AI), where functional language is often used to describe complex systems.
The Case: Payment Handler as Functional Language
The case, involving a dispute over software patents related to payment processing technologies, centered on the term “payment handler.” The court examined whether this term invoked means-plus-function claiming under § 112 ¶6, which applies when a claim term is expressed in purely functional terms, without reciting sufficient structural detail. Under this provision, if a claim lacks structural detail, it is considered indefinite unless the specification provides enough supporting structure or an algorithm corresponding to the claimed function.
The Federal Circuit began by discussing whether the term “payment handler” overcame the presumption against invoking means-plus-function claiming. The court ruled that the term indeed lacked sufficient structural specificity, as it only described the function of handling payments without specifying how this was achieved. The court likened the term “payment handler” to “module”, which has previously been considered a nonce term in patent law, representing a vague description of a software or hardware component that performs a specified function.
Why “Payment Handler” Was Deemed Indefinite
The court rejected several arguments put forth by the patent holder. For one, the plaintiff argued that terms like “operable to,” “configured to,” and “that” used in the claim language conferred sufficient structure to avoid means-plus-function treatment. The Federal Circuit noted that while these terms are often used in structural contexts, they do not automatically prevent means-plus-function claiming. Citing the case Rain Computing, Inc. v. Samsung Electronics America, the court pointed out that the applicability of § 112 ¶6 depends on the specific context and nature of the claims.
The court also addressed the argument that the “payment handler” terms were part of a recognized class of software structures like “code” or “applications,” which the court in Dyfan, LLC v. Target Corp. found to be sufficiently structural. However, the Federal Circuit emphasized that, unlike “code” or “application,” the term “payment handler” had no established meaning within the software development community. The patent holder had failed to provide expert testimony or concrete examples showing how the term conveyed structure.
Additionally, the court rejected the argument that the surrounding claim language—such as defining inputs, outputs, and operation of the payment handler—provided enough detail to make the term structural. The claim did not explain how the payment handler functioned, nor did it outline the specific “rules” or algorithm that would govern its operation. The Federal Circuit noted that the specification of the patent simply repeated the claim language without offering any substantial details about the underlying structure of the payment handler.
In essence, the court concluded that the term “payment handler” was functionally indefinite and did not include the necessary structural disclosure to satisfy § 112 ¶6. As a result, the court invalidated the patent claims that relied on this vague term.
Implications for AI Patent Applications
Although this decision did not directly address artificial intelligence (AI), it offers significant insights for AI-related patent drafting, where functional terms are often used to describe complex technologies. AI inventions, particularly those involving machine learning models, neural networks, and other advanced algorithms, may face similar challenges when their claims rely heavily on functional descriptions.
In AI patents, terms like “classifier,” “predictive model,” or “neural network” are often used to describe the operations of a system without fully detailing the underlying algorithm or architecture. While these terms may be widely accepted in the field, patent drafters must be cautious when they lack sufficient structural disclosure in the specification.
Provide Detailed Structural Descriptions: Instead of relying on broad, functional terms like “classifier” or “model,” drafters should disclose as much structural detail as possible, including algorithms and specific AI techniques used. For example, terms like “feed-forward neural network,” “convolutional neural network,” or “generative pre-trained transformer” provide concrete examples of structures and algorithms that could support the claims and avoid indefiniteness challenges.
Avoid Ambiguous Terminology: Terms like “handler” or “module,” which are commonly used as placeholders for functional components, should be avoided or supplemented with detailed explanations of their structure and operation. If a term like “payment handler” is essential, ensure the patent specification includes an in-depth description of the specific software or hardware involved and how it performs its function.
Use Recognized AI Terms for Structure: Where possible, use terms that are already well understood to connote structure in the AI field. For instance, the term “model” could be more structural in the AI context than terms like “classifier,” especially when it is described with reference to specific AI architectures and algorithms.
Include Dependent Claims for Clarity: Dependent claims can be used to provide more specific details on the structure of AI systems, such as the type of neural network or the algorithm being used.
Don’t Rely Solely on Claim Language: As the court emphasized, merely parroting the claim language in the specification is not enough. It’s crucial to explain the structural components in detail, particularly for AI inventions that involve complex algorithms and system architectures.
Conclusion: The Need for Clarity in AI Patents
The Federal Circuit’s decision underscores the importance of clarity and specificity in patent claims, particularly in fields like software and AI, where functional terms are commonly used. Patent applicants must ensure that functional language is supported by concrete structural details to avoid claims being deemed indefinite under § 112 ¶6. By providing comprehensive descriptions of the structure and algorithms underlying their inventions, AI patent drafters can strengthen their patent applications and reduce the risk of invalidation due to indefiniteness.
As AI technologies continue to evolve, patent law will need to adapt, and the case serves as a timely reminder that functional claims must be backed by sufficient structure to withstand legal scrutiny.