The two-step patent eligibility test from Alice Corp. v. CLS Bank International asks, first, whether a patent claim is directed to an abstract idea, a law of nature, or a natural phenomenon, and second, if it is, whether the claim adds an inventive concept that transforms the exception into something patent-eligible. The Supreme Court laid out this framework unanimously in 2014, and it now governs every software patent application and challenge in the United States. Running a familiar business idea on a generic computer, the Court held, does not clear the bar.1Justia. Alice Corp. v. CLS Bank Intl, 573 U.S. 208 (2014)
The Dispute Behind the Test
Alice Corporation held four patents on a computerized system for managing settlement risk in financial transactions. The idea was to use a computer as a neutral third-party intermediary that tracked accounts and released a trade only when both sides had the assets to complete it. The patents included method claims (the steps of the process), system claims (computer hardware configured to run those steps), and media claims (software code stored on a computer-readable medium).
CLS Bank International, which runs a global currency-transaction network, sued for a declaration that the patents were invalid, arguing they described an old financial practice dressed in computer language. The Federal Circuit split badly on the reasoning, and the Supreme Court took the case to settle the standard.
The Court’s Holding
Justice Thomas wrote for a unanimous Court. Alice’s claims, he held, were “drawn to the abstract idea of intermediated settlement, and merely requiring generic computer implementation fails to transform that abstract idea into a patent-eligible invention.”1Justia. Alice Corp. v. CLS Bank Intl, 573 U.S. 208 (2014) Every category of claim fell together. The method claims recited the abstract idea on a generic computer. The system claims described “a handful of generic computer components configured to implement the same idea.” The media claims added nothing beyond the other two.
The individual computer functions Alice pointed to, such as creating shadow accounts, obtaining data, adjusting balances, and issuing automated instructions, were all “purely conventional.” Taken together, they amounted to the concept of intermediated settlement performed by a computer, with no improvement to the computer itself and no advance in any other technology.1Justia. Alice Corp. v. CLS Bank Intl, 573 U.S. 208 (2014)
Step One: Is the Claim Directed to a Judicial Exception?
The first question is whether the claim is directed to one of three off-limits categories: abstract ideas, laws of nature, or natural phenomena. The Court calls these “the basic tools of scientific and technological work,” and the concern is preemption. A patent that locks up a fundamental concept blocks everyone else from building on it in ways the patent holder never imagined.2United States Patent and Trademark Office. MPEP 2106 – Patent Subject Matter Eligibility
Alice’s concept of intermediated settlement, using a neutral third party to manage a transaction, was “a fundamental economic practice long prevalent in our system of commerce.”1Justia. Alice Corp. v. CLS Bank Intl, 573 U.S. 208 (2014) That made it an abstract idea. The Court drew a direct line to Bilski v. Kappos, where a method of hedging risk in energy markets had already been held unpatentable for the same reason.
Step Two: Does the Claim Add an Inventive Concept?
If step one identifies a judicial exception, the analysis moves to whether the claim adds something “significantly more.” A genuine inventive concept is required. Reciting the idea and then saying “apply it on a computer” is not enough. What counts is a specific technical improvement: a new way of managing memory, a faster encryption method, a more efficient data structure.2United States Patent and Trademark Office. MPEP 2106 – Patent Subject Matter Eligibility
The test looks at each claim element on its own and at the combination as a whole, asking whether it goes beyond “well-understood, routine, and conventional” activity in the field. If the computer is just doing what a human already does with pen and paper, tracking balances, comparing numbers, flagging discrepancies, there is no inventive concept. Alice’s patents failed at step two because every step the computer performed was something accountants and financial intermediaries had been doing manually for centuries.1Justia. Alice Corp. v. CLS Bank Intl, 573 U.S. 208 (2014)
The framework itself was not invented in Alice. The Court had first outlined it in Mayo Collaborative Services v. Prometheus Laboratories, Inc., a case about medical diagnostic patents, which is why practitioners now call it the Alice/Mayo test.3Legal Information Institute. Mayo Collaborative Services v. Prometheus Laboratories, Inc. Alice extended it decisively to software.4Office of the Law Revision Counsel. 35 U.S.C. 101 – Inventions Patentable
How the USPTO Applies the Two Steps
Patent examiners work through a version of the framework that splits step one into two prongs. The distinction matters because a claim can pass at prong two without needing to reach step two at all.
Prong One: Does the Claim Recite a Judicial Exception?
The examiner first checks whether the claim language sets forth an abstract idea, law of nature, or natural phenomenon. To keep this consistent, the USPTO groups abstract ideas into three categories: mathematical concepts (formulas, equations, calculations); certain methods of organizing human activity (fundamental economic practices like hedging or insurance, commercial interactions like contracts and marketing, and managing personal behavior); and mental processes (observations, evaluations, judgments, or opinions a person could perform in their head). Alice’s intermediated settlement claims fell squarely into the second category. A claim that does not fit any of the groupings is treated as eligible without further analysis.2United States Patent and Trademark Office. MPEP 2106 – Patent Subject Matter Eligibility
Prong Two: Is the Exception Integrated into a Practical Application?
If the claim does recite an exception, the examiner asks whether the claim as a whole integrates it into a practical application. Several factors help answer that: whether the claim improves how a computer works or improves another technical field; whether the exception is tied to a particular machine that is integral to the claim rather than tacked on; whether the claim transforms an item into a different state or thing; and whether the exception is applied in a meaningful way, not just linked to a general technological environment.2United States Patent and Trademark Office. MPEP 2106 – Patent Subject Matter Eligibility
Certain shortcuts reliably fail. Saying “apply it on a computer,” adding insignificant extra steps like generic data gathering, or narrowing the idea to a broad field of use such as “on the internet” will not integrate anything. A claim that clears prong two is eligible, and the examiner stops there.
Step 2B: Significantly More
Claims that fail both prongs of step one face the last question, which tracks Alice’s step two directly: do the additional elements amount to significantly more than the judicial exception? Here the examiner does consider whether those elements are well-understood, routine, and conventional in the field. A claim reciting only standard computer operations will not survive.2United States Patent and Trademark Office. MPEP 2106 – Patent Subject Matter Eligibility
What Kinds of Software Claims Survive
Alice did not end software patents. It shifted the ground: a claim aimed at a specific technical improvement still gets through, while a claim aimed at a broad business concept usually does not. Two Federal Circuit decisions mark the line.
In Enfish, LLC v. Microsoft Corp. (2016), the court upheld claims covering a self-referential database table. “The plain focus of the claims,” the court wrote, “is on an improvement to computer functionality itself, not on economic or other tasks for which a computer is used in its ordinary capacity.” The self-referential table produced concrete benefits: increased flexibility, faster search times, and smaller memory requirements than conventional structures. Much of computer progress, the court noted, consists of software improvements “that, by their very nature, may not be defined by particular physical features but rather by logical structures and processes,” and nothing in Alice excludes that entire field.5United States Court of Appeals for the Federal Circuit. Enfish, LLC v. Microsoft Corporation
In McRO, Inc. v. Bandai Namco Games America Inc. (2016), the court upheld claims covering an automated method for synchronizing animated lip movements to speech. The claims used specific rules evaluating sub-sequences of phonemes to determine how animated faces should move. Those were not just “rules” in the abstract but “rules with specific characteristics” that improved a technological process previously done by human animators. The automation did not copy what humans did onto a computer; it replaced a subjective artistic process with a defined, rule-based technical method.6United States Court of Appeals for the Federal Circuit. McRO, Inc. v. Bandai Namco Games America Inc.
The USPTO extended the same reasoning to artificial intelligence in July 2024, publishing updated guidance and new examination examples. Whether an invention was made with the help of AI has no bearing on eligibility. The new examples treat as eligible claims to neural networks for detecting malicious network packets and AI-based speech separation systems, because each described specific technical implementations (particular network architectures, defined training algorithms, concrete real-world applications) rather than the abstract concept of “using AI” to solve a problem.7Federal Register. 2024 Guidance Update on Patent Subject Matter Eligibility, Including on Artificial Intelligence
Drafting Claims That Pass the Test
Knowing the framework is one thing; writing claims that clear it is another. The specification (the detailed written description of the invention) does most of the work. For a computer-implemented invention, the USPTO requires the specification to describe the algorithm that performs the claimed function in sufficient detail. Naming a function without disclosing how it is carried out is not enough. The algorithm can appear as a flowchart, pseudocode, a mathematical expression, or prose, but it has to be there.8United States Patent and Trademark Office. MPEP Section 2161 – Three Separate Requirements for Specification Under 35 U.S.C. 112(a)
A few habits improve a claim’s chances:
- Frame the claim around a technical problem and a defined solution. “A method of reducing fraud” reads like an abstract idea; the specific data-processing steps, comparisons, and thresholds that identify fraudulent patterns read like a technical solution. Enfish and McRO both turned on that framing.
- Describe how the software improves the functioning of the computer, and back it up with benchmarks or comparisons to prior approaches. Improvements to computer functioning satisfy the practical-application inquiry at prong two.2United States Patent and Trademark Office. MPEP 2106 – Patent Subject Matter Eligibility
- Avoid purely functional language. “A system for optimizing delivery routes” invites an abstract-idea rejection. “A system that applies a modified Dijkstra algorithm to real-time GPS data weighted by historical traffic density” describes something concrete.
- Include both the logical process and the hardware environment. The specification should convey that something was built, not that an idea was had and someone else could code it.
Detail scales with complexity. For a straightforward database query, a brief description of the query logic may suffice. For a machine-learning model, the specification should cover the training data, network architecture, loss function, and how the model’s output is applied to a real-world task.8United States Patent and Trademark Office. MPEP Section 2161 – Three Separate Requirements for Specification Under 35 U.S.C. 112(a)
Where Reform Stands
The framework has drawn steady criticism. Some patent holders argue the abstract-idea inquiry is too vague and applied inconsistently; others say it correctly removes low-quality patents that never should have issued. The Patent Eligibility Restoration Act of 2025 (H.R. 3152) would replace the Alice/Mayo framework with a statutory list of exclusions. Under the proposed language, a patent could not cover a standalone mathematical formula, a process that is “substantially economic, financial, business, social, cultural, or artistic,” a purely mental process, an unmodified human gene, or an unmodified natural material. The bill carves out a significant exception: a claimed invention is not excluded if it “cannot practically be performed without the use of a machine or manufacture.” It would also bar courts from discounting or disregarding any claim element in the eligibility analysis.9Congress.gov. Text – H.R.3152 – 119th Congress (2025-2026) – Patent Eligibility Restoration Act of 2025 As of mid-2026, the bill has not become law. The Alice/Mayo two-step test remains the governing standard.