Is software patentable in America?

 The US patent system for software is vastly different from Australia's new, more pragmatic approach, due to the controlling influence of a 2014 Supreme Court decision.

The law in America is highly challenging, but ultimately, software is patentable if the application is drafted to be a technical innovation, not merely a business idea.

Patenting Software in America: Navigating the Post-Alice Landscape

For Australian innovators looking to protect their software and AI inventions in the United States, the legal landscape is both the most rewarding and the most challenging in the world. While nearly two-thirds of all US utility patents granted involve software in some form, the path to a valid patent is a narrow one, dictated by the Supreme Court’s 2014 decision in Alice Corp. v. CLS Bank International.

The US test demands that your invention be more than a brilliant business concept; it must be a demonstrable technical solution to a technological problem.

The Two-Step Alice Test for Patent Eligibility

US patent law (35 U.S.C. § 101) excludes Abstract Ideas, Laws of Nature, and Natural Phenomena from patent eligibility. For software, the challenge is that courts and the US Patent and Trademark Office (USPTO) almost always treat the claims as being directed to an Abstract Idea.

The two-part test, derived from Alice and Mayo, is the exclusive method used to determine eligibility:

Step 1: Is the claim directed to a judicial exception?

In the context of software and business methods, the claims are analyzed to see if they are "directed to" a recognized abstract idea, such as:

A fundamental economic practice (like intermediated settlement or a financial transaction).

A method of organizing human activity (like rules for a game or managing an online transaction).

A mathematical formula or mental process (like comparing data or performing basic calculations).

For most business-related software, the answer to Step 1 is almost always "Yes," which leads directly to the more critical Step 2.

Step 2: Does the claim contain an "inventive concept"?

If the claim is directed to an abstract idea, it must contain an "inventive concept" sufficient to transform the abstract idea into a patent-eligible application.

This inventive concept must be "significantly more" than the abstract idea itself and cannot be composed merely of "well-understood, routine, or conventional" activities previously known to the industry.

To satisfy Step 2, the claims must show one of the following:

Improvement to Computer Functionality: The software must solve a problem in the technical field of computing itself.

Examples: Speeding up data retrieval, reducing memory usage, or enabling computations that were previously unavailable.

Unconventional Technical Solution: The claims must solve a problem that is "necessarily rooted in computer technology" (i.e., not a problem that exists in a traditional, non-computer context) and do so in an unconventional way.

Example (Win): In DDR Holdings v. Hotels.com, a method for creating a "hybrid" webpage was found eligible because it solved a problem specific to the Internet—retaining a user on a host site after clicking a third-party link—with a specific, unconventional technical solution.

New Combination of Known Elements: An inventive concept can sometimes be found in the non-conventional and non-generic arrangement of known, conventional pieces. The ordered combination must achieve a new technical result.

What is NOT enough: Merely adding "on a computer" to an abstract idea, or using a computer to perform routine data gathering or processing steps, will fail Step 2.

The Latest Challenge: AI and "Do It With AI" Claims

The battleground has recently shifted to Artificial Intelligence (AI) and Machine Learning (ML).

In the landmark 2025 decision Recentive Analytics, Inc. v. Fox Corp., the Federal Circuit held that claims that "do no more than apply existing methods of machine learning to a new data environment are patent ineligible."

This means that simply taking an abstract idea (like "producing network maps" or "predictive analytics") and saying "do it with AI" or "do it with ML" is not enough to constitute an inventive concept. To be eligible, an AI/ML invention must offer a concrete technical innovation that goes beyond using "off-the-shelf” algorithms.

How to Succeed: Lessons for Australian Applicants

My own personal trade mark conflict program that I invented in 2011 (that automatically detected whether any new trade mark application conflicting with any one of thousands of my client’s prior registration) was successfully patented in Australia because it achieved a result impossible for a human.  This is a perfect example of the mindset needed for the US:

Key Strategy	How it Addresses the Alice Test
Focus on the "HOW"	Do not claim the business outcome (the what). Instead, claim the specific technical steps and architecture that implement the idea in a non-generic way.
Specify Technical Solution	Claims should detail a specific technical solution to a technological problem, or an improvement to the computer's operation (e.g., a specific, non-conventional algorithm for data compression or searching).
Avoid Generic Language	Never use generic phrases like "processor," "database," or "computer-readable medium" if those elements are simply performing their routine functions. Instead, claim the computer configured to perform the specific, novel tasks.
Conflate Sections 101, 102, 103	While legally separate, in practice, patent eligibility (Section 101) is often determined by whether the claimed elements are non-conventional (which is a novelty/inventive step inquiry). Claims that are genuinely novel and non-obvious have a far better chance of passing the Alice test.

The path to patenting software in the US is narrow and challenging, but for truly inventive solutions that are implemented technically, it is still very much open.