Software Patents and the Alice Test: US Patent Eligibility Explained

Why software patents are contested in the US

Software and AI businesses often assume that a working, novel system is automatically patentable. In the United States the harder question comes earlier, at eligibility. Section 101 of the Patent Act (35 USC 101) sets out what kinds of subject matter can be patented at all, and the courts have layered a set of judicial exceptions on top of it. Abstract ideas, laws of nature and natural phenomena sit outside the system, however clever or useful the implementation. Many software claims fall foul of the abstract-idea exception, which is why eligibility, rather than novelty or obviousness, is frequently where a software application succeeds or fails. Eligibility is a separate question from whether the invention is new or non-obvious, and a claim must clear all of those hurdles. This overview sits under our broader guide to US patents, which covers the wider filing picture.

The Alice/CLS Bank two-step test

The framework that governs software eligibility comes from the Supreme Court decision in Alice Corp v CLS Bank, building on the earlier Mayo case, so practitioners often call it the Alice/Mayo test. It governs subject-matter eligibility generally, from diagnostics to natural products, but it is where most software claims are decided. It runs in two steps.

Step one asks whether the claim is "directed to" a patent-ineligible concept, most commonly an abstract idea. The USPTO's eligibility guidance sorts abstract ideas into three groupings: mathematical concepts, certain methods of organising human activity (which is where fundamental economic practices sit), and mental processes. If the claim is not directed to one of these, it is eligible under the Alice/Mayo exception analysis, though it must still satisfy the separate requirements of novelty, non-obviousness and adequate written description.

Step two applies only if step one is answered against the applicant. It asks whether the claim contains an "inventive concept", something that transforms the abstract idea into a patent-eligible application by adding "significantly more" than the idea itself. Reciting the abstract idea and then saying "do it on a computer" is not enough. Generic computer components performing their ordinary functions do not supply the inventive concept the test is looking for.

In practice the USPTO applies a modified version of this framework during examination. Its Step 2A first asks whether the claim recites one of the exceptions, then whether the exception is nonetheless integrated into a practical application; only if the answer is no does the examiner reach Step 2B, the inventive-concept inquiry. So a claim reciting an abstract idea can still be found eligible without ever reaching the "significantly more" stage, which is why how the claim ties the idea to a practical application matters so much.

The abstract-idea problem for software and AI

The recurring failure mode is a claim that describes a result or a business outcome and leaves the generic computer to achieve it. Taking a known practice, automating it, and claiming the automation on standard hardware tends to be treated as an abstract idea implemented on a generic computer, which does not survive the two steps. AI inventions raise the same tension in a sharper form, because a claim written around a model that "predicts", "classifies" or "optimises" can read as a mathematical concept or a mental process dressed in software. For a wider view of how eligibility interacts with machine learning and generative systems across jurisdictions, see our cross-border note on AI and IP.

The US Patent and Trademark Office issues its own eligibility guidance for examiners, and it evolves. Case law continues to move as well. Because outcomes are fact-specific and the guidance is periodically updated, you should not treat any single decision as a fixed rule; confirm the current position with the USPTO or with qualified US patent counsel before relying on it.

Drafting to a concrete technical improvement

The practical response to the abstract-idea problem is to anchor the invention in a concrete technical improvement rather than an abstract result. Claims have fared better when they describe how the computer itself, or a technical process, is made to work better: a more efficient use of memory or bandwidth, a reduction in processing steps, an improvement to how a network, database or interface functions, or a specific technical mechanism that could not be performed by a person with pen and paper. That kind of improvement is also what tends to satisfy the practical-application question at the USPTO's Step 2A. The improvement should be visible in the claim language and supported in the written description, not asserted only in the marketing.

For AI inventions the same logic applies. Tying the claim to a specific technical implementation, a particular architecture, a training or inference technique that solves a defined engineering problem, or a measurable improvement in the functioning of a system, gives the eligibility analysis something concrete to hold onto. Vague functional claiming, by contrast, invites an abstract-idea rejection. Getting this right at the drafting stage is far cheaper than trying to rescue an application later, and it is genuinely difficult to do well, which is why experienced US patent counsel earns their fee here.

On costs and timing more generally, official USPTO fees apply and the examination timeline varies; confirm the current amounts and typical periods with the USPTO or local counsel rather than relying on a figure quoted online.

Where IPEnvoy fits

IPEnvoy is not a law firm and does not provide legal advice. This is general information to help you understand the landscape, not a substitute for a professional opinion on your invention; confirm the current position with the USPTO and a qualified local IP professional. Eligibility calls under Alice are nuanced and turn on the specifics of your technology and how the claims are written, which is exactly the kind of work that benefits from a specialist. If you would like, IPEnvoy can introduce you to vetted US patent attorneys who handle software and AI subject matter, so you can get your eligibility position assessed and your claims drafted with the two-step test in mind from the start.