Key Takeaways for AI/Machine Learning Patent Filers - Recentive Analytics, Inc. v. Fox Corp

In a recent US Federal Circuit decision (Recentive Analytics, Inc. v. Fox Corp), the US court of appeals considered the validity of patent claims aimed at using machine learning for tasks such as event scheduling and network mapping. The court ruled that merely applying generic machine learning techniques to a conventional data processing problem does not meet the patent eligibility threshold established under 35 U.S.C. § 101. In essence, the claimed methods were deemed too abstract, lacking a specific inventive concept that transforms standard data processing into something patentable.

The court applied the two-step test from the Alice decision to machine learning, and has likely set a precedent going forward where the Alice test may be used to construe not only computer implemented software inventions, but also inventions relying on machine learning in some way. This suggests that use of generic machine learning technology is now considered a conventional technique, in the same way that using a generic computer to perform a task is also deemed “conventional”, and therefore not sufficient to achieve patentability.

Interestingly, the decision specifically highlights a description passage in the patent specification, where the patented method employs “any suitable machine learning technique”. Whether this term resulted in the machine learning technique being deemed generic is unclear, however more detail of the machine learning technology within the specification may have resulted in a different conclusion.

It is likely this outcome would be replicated in many other key jurisdictions, and serves as a reminder to focus on concrete, nonobvious technical improvements when attempting to patent machine learning and AI technology.

For inventors looking to secure AI patents in the United States and elsewhere, this case highlights the importance of demonstrating a clear technical advancement – not just a routine implementation of known algorithms or techniques via machine learning/AI. Innovators should clearly articulate how their invention offers a novel solution to a technical problem, and ensure this is supported by sufficient technical details in the patent specification of how this technical problem is solved.