Introduction
As artificial intelligence (AI) technologies continue to advance at breakneck speed, a recent dispute over the patentability of AI-powered inventions has thrust U.S. patent law into the spotlight. In a contentious appeal before the U.S. Court of Appeals for the Federal Circuit, software developer XYZ Innovations warns that the court’s recent application of Section 101 of the Patent Act has pushed patent jurisprudence to its “breaking point.” With billions of dollars riding on the ability to protect AI breakthroughs, the outcome could reshape how—and whether—innovators secure legal safeguards for their creations.
Background: Patent Eligibility Under Section 101
• 35 U.S.C. § 101 sets the boundaries for patentable subject matter in the United States.
• Since the Supreme Court’s 2014 Alice decision, courts have applied a two-step test to determine when claims are directed to ineligible “abstract ideas.”
– Step 1: Does the claim target an abstract idea, law of nature, or natural phenomenon?
– Step 2: If so, does it add an “inventive concept” that transforms the abstract idea into a patent-eligible application?
• Critics say Alice has sown confusion: courts struggle to identify what constitutes an “abstract idea” and what qualifies as an “inventive concept,” particularly in software and AI cases.
• As a result, many AI-related patents have been invalidated at the district-court and Patent Trial and Appeal Board levels.
The AI Patent Dispute: PromptTech v. LargeAI
• PromptTech, Inc. sued LargeAI Corp. in the U.S. District Court for the District of Delaware, accusing LargeAI of infringing two patents covering methods for training and deploying generative-language AI models across distributed computing nodes.
• The patents describe a pipeline that preprocesses user inputs, runs a transformer-based neural network in parallel, then postprocesses results to maintain data provenance and compliance.
• LargeAI moved for summary judgment of invalidity under § 101, arguing that the claims merely recite abstract data-processing steps on generic hardware.
• In early 2024, Judge Sue A. Bolton granted the motion, holding that the asserted claims target abstract ideas with no “inventive concept” beyond implementing them on commonplace servers.
• On appeal, a three-judge Federal Circuit panel affirmed, applying Alice and finding the claims ineligible because “any skilled practitioner could perform the claimed steps mentally or with pencil and paper, then transfer them to a computer.”
• Separately, the Patent Trial and Appeal Board invalidated the same patents in inter partes review, citing similar reasoning.
XYZ Innovations’ Amicus Brief
• XYZ Innovations, a midsize AI software developer, filed an amicus brief urging the Federal Circuit to rehear the case en banc—or to spur Supreme Court review.
• Key arguments:
– “Breaking Point” for Patent Law: The panel’s broad reading of Alice sweeps in virtually all AI innovations, making it nearly impossible to predict which AI inventions are protectable.
– Undermining Congressional Intent: Congress intended Section 101 to have “wide scope,” yet courts now treat nearly any software-implemented invention as an abstract idea.
– Erosion of Investment Incentives: Vague eligibility rules discourage venture capital and corporate R&D in AI, as investors fear that patents will be invalidated post-grant.
• XYZ proposes restoring the Federal Circuit’s pre-Alice guidance, including:
– A clear “technical improvement” criterion, focused on whether the claim improves computer functionality itself.
– Recognition that machine-implemented processes—even if they build on abstract ideas—can be eligible if they yield concrete, real-world benefits.
– Emphasis on claim language and context, rather than deeming any algorithm “abstract” based on high-level descriptions.
Broader Implications and Next Steps
• The Federal Circuit’s upcoming decision on rehearing could recalibrate the Alice framework, clarifying whether and how AI inventions qualify for patent protection.
• A refusal to grant en banc review would leave the current panel ruling in place, potentially deepening uncertainty and prompting more appeals to the Supreme Court.
• Lawmakers and industry groups are monitoring this case as a bellwether for U.S. competitiveness in AI. If AI patent eligibility remains unsettled, some companies may shift key R&D overseas or rely more heavily on trade-secret protection.
• Beyond PromptTech, hundreds of AI-related cases are pending in district courts and before the Patent Office. A clear standard is critical to guiding patent applicants, examiners, and judges alike.
Three Takeaways
1. Ongoing Confusion: Courts applying Alice continue to struggle with defining “abstract ideas” and “inventive concepts,” especially in AI and software contexts.
2. Industry Alarm: An amicus brief by XYZ Innovations warns that unclear patent rules are deterring investment and threatening U.S. leadership in AI.
3. Stakes Are High: The Federal Circuit’s en banc decision—or subsequent Supreme Court review—could reshape eligibility standards for all AI-powered inventions.
Frequently Asked Questions
Q1: What makes an invention patent eligible under current law?
A1: Under Section 101, an invention must fall into one of four statutory categories (process, machine, manufacture, or composition of matter) and must not be directed to an abstract idea, law of nature, or natural phenomenon without an “inventive concept” that transforms it into a practical application.
Q2: Why are AI patents often invalidated today?
A2: Many AI patents claim algorithms or data-processing techniques on generic hardware. Courts often deem these to be abstract ideas implemented on a computer, lacking the additional inventive features required by the Alice test.
Q3: What could change as a result of this case?
A3: If the Federal Circuit grants en banc review and refines Section 101 guidance, it may introduce a clearer “technical improvement” test or other criteria that make it easier to protect certain AI inventions. A denial of review could drive the issue to the Supreme Court for ultimate resolution.
