Artificial intelligence is being used in nearly every industry, from scheduling and advertising to fintech, healthcare, manufacturing, and software automation. But using AI does not automatically make an invention patentable.
The Federal Circuit’s decision in Recentive Analytics, Inc. v. Fox Corp. is now one of the most important cases for AI and machine-learning patent eligibility. The case involved patents directed to using machine learning to generate network maps and schedules for television broadcasts and live events. The Federal Circuit held that the claims were patent ineligible because they merely applied generic machine-learning techniques to a particular environment, without claiming an improvement to the machine-learning technology itself.
The court also explained that iterative training and real-time updating are inherent features of machine learning. In other words, simply saying that a model is trained, updated, or adjusted based on new data may not be enough to make an AI-related invention patent eligible.
The significance of Recentive has grown quickly. The Shepard’s report shows no negative subsequent appellate history, notes that the Supreme Court denied certiorari on December 8, 2025, and identifies 176 citing decisions. Courts and the PTAB are now using Recentive in cases involving AI, neural networks, predictive models, optimization systems, image analysis, fraud detection, advertising technology, and other software-based inventions.
The main lesson is simple: an AI patent application should not merely claim the use of AI to produce a useful result. It should explain the technical improvement.
For example, a weak disclosure might say:
The system uses machine learning to analyze data and generate an optimized recommendation.
A stronger disclosure would explain what data is used, how the data is transformed, how the model is trained or structured, what technical problem is being solved, and how the claimed approach improves computer functionality, model operation, data processing, image analysis, resource usage, or another technical process.
Courts are also using Recentive to reject “new field of use” arguments. Applying machine learning to a new industry, such as advertising, scheduling, fintech, healthcare administration, or business analytics, may not be enough if the claims only use conventional AI tools in a generic way.
That does not mean AI inventions cannot be patented. Some courts have distinguished Recentive where the claims included specific technical detail or were directed to an improvement in image processing, risk analysis, model operation, or another technical area.
The takeaway for inventors is that AI-related patent applications need to be drafted carefully. The application should identify the technical problem, explain the technical solution, and make sure the claims reflect the actual improvement — not just the desired result.
AI can still be part of a patentable invention. But after Recentive, a patent application needs to do more than say “use machine learning.” It needs to explain what the technology does differently and why that difference matters.
