In April, the Federal Circuit issued a ruling on key patent laws, including artificial intelligence. in Lyromive Analytics, Inc. v. FoxCorpthe court addressed a central question facing many AI-driven companies. When are solutions that apply machine learning to real-world problems original and patentable? The Federal Circuit confirmed the court's rejection of the underlying lawsuit at the appeal stage under §101 and determined that it aims to meet a knowledgeable standard under 35 USC §101 without disclosing technological advances to underlying machine learning technologies.
Last Wednesday, Redive filed a total petition for panel rehearsals and rehearsals en banc The April decision erased the line between §101 and §102/103, claiming that the §112 activation requirement was incorrectly imported. The Rimaibu petition also points to the fact that several judges in the Federal Circuit previously urged the court to consider the boundaries between eligibility and these validity requirements.
Below we analyze the April ruling regarding dismissal under §101. It also explains the impact of the Federal Circuit denying a rehearsal request for redives on companies developing AI technology.
Decided on the Federal Circuit in April 2025
Stramive Analytics, Inc. used machine learning to schedule live events and claimed four patents related to TV program allocation. The patents were divided into two groups. (1) The “Machine Learning Training” patent focuses on generating optimized schedules for live events. (2) The “Network Map” patent focuses on generating “Network Maps” that assign programming to the television market to maximize viewership ratings.
All claims include collecting data, applying machine learning algorithms, and generating output. A schedule or programming map. The patent is a standard machine learning technology (for exampleregression, neural networks, decision trees, etc.) to practice your invention. Importantly, I have recently acknowledged that patents do not advocate for machine learning techniques themselves, but rather apply known machine learning techniques to specific scenarios.
The court found that claims were simply directed towards the abstract idea of optimizing scheduling and programming using machine learning, and that claims lacked original concepts.
Recent claims have argued that some are actually directed towards improving technology, as the claims require repeated training of machine learning models. However, the court rejected the argument as it found that iterative training was embedded in the machine learning process.
He also argued that recent application of machine learning is not common as the algorithms work dynamically and created a way for maps and schedules to be automatically customizable and updated with real-time data. However, the court emphasized that neither the claim nor the specifications disclosed such improvements.
Instead, the claims simply applied common computing techniques (machine learning) to specific use cases, generating network maps and schedules. And the courts have long recognized that abstract ideas are not abstract simply by limiting their field of use. Furthermore, the court explained that, in a general matter, the patent described the use of already available technology as a tool to carry out the claimed process as an abstract.
Finally, the court has found that the clear, original concept of Remerger is “using machine learning to dynamically generate optimized maps and schedules based on real-time data and update them based on changing conditions.” Therefore, there was no concrete, original concept that provided qualifications for claims.
AI Patent Caution Signal
The April ruling strengthens the standard for patenting AI-related inventions, especially when it relies on ready-made ML models. If the April ruling is upheld, how the April ruling will affect practitioners and innovators is as follows:
“New methods using AI” are not sufficient. Simply assert the process Uses If the underlying method is traditional and ML applications are common, machine learning does not grant patent eligibility either repeatedly or dynamically.
The claim should indicate a technical improvement. To pass §101, an AI claim must demonstrate certain technological advances. Practitioners should try to understand the problems that need to be resolved. It's just business goals. Rather than simply asking how AI fixes business problems, practitioners need to focus on the technical hurdles of implementing and incorporating AI to solve technical problems.
Reading between lines means that under April, this kind of claim will be awarded.
- New ML training methods
- New ML Data Architecture or Processing Method
- Improved performance for computing systems that run ML algorithms
Functional outcomes do not save abstract claims – the April ruling shows that the panel remains skeptical of claims that only recite high-level outcomes (for example“maximize evaluations”, “optimize schedules”) without explaining
how These results are achieved in a non-abstract and original way. However, whether the claim will recite whether rehearsals are permitted or not how When it comes to AI patents, it should be a prolonged question in the minds of all practitioners.
A recent rehearsal petition argues that under the April decision, the application of existing machine learning models is not patent qualifying and, as a result, does not curb investment in the invention of machine learning that changes the world. It remains to be seen whether the Federal Circuit will grant the rehearsal petition. However, given the interests and issues in play, this is certainly a case worth looking at carefully. stay tuned!
AI, algorithms and abstract ideas: The Federal Circuit, recent v. Strengthen FOX restrictions
The content of this article is intended to provide a general guide to the subject matter. You should seek expert advice on your particular situation.
