Recentive v. Fox: The AI Eligibility Ceiling
Federal Circuit, decided 18 April 2025. Read from the decision; quotations verified against its text.
In the first appellate decision to squarely address the question, the Federal Circuit held on April 18, 2025 that patent claims doing no more than applying established machine learning techniques to a new field are directed to an abstract idea and ineligible under 35 U.S.C. § 101, while expressly leaving room for claims that improve the machine learning itself.
Recentive patented the use of machine learning to build television broadcast schedules and network maps. The Federal Circuit held the patents ineligible, calling this a question of first impression and answering it directly: claims that do no more than apply established machine learning methods to a new data environment are abstract ideas.
Iterative training and real time updating did not save them, because the court found those are ordinary attributes of machine learning rather than improvements to it.
The decision does not say machine learning is unpatentable. It says the improvement has to be to the technology, and it has to appear in the claims. The opinion's own final paragraph leaves room for claims that may lead to patent-eligible improvements in machine learning.
What This Case Is Not
- It does not hold that machine learning inventions are ineligible as a class.
- It does not tell you whether your claims recite an improvement. That is a legal conclusion a patent attorney reaches on the actual claims.
- It does not make any drafting pattern sufficient. Reciting an improvement in the specification is not the same as reciting it in the claims.
Educational, not legal advice. Whether any particular claim is eligible under § 101 is a legal conclusion a qualified patent attorney reaches on the actual claims.
Sources
- Recentive Analytics, Inc. v. Fox Corp., No. 23-2437 (Fed. Cir. Apr. 18, 2025)
- Software patents after Alice: the wider picture