In a recent opinion from the United States Court of Appeals for the Federal Circuit, the federal appeals court that handles all patent litigation appeals in the U.S., the Court invalidated four (4) patents that utilized machine learning to solve a problem in the television broadcasting industry.
The Plaintiff, Recentive Analytics, Inc., owned four (4) patents that utilized machine learning to address a problem faced by television broadcasters and similar entities—namely, how to optimize the scheduling of live events and “network maps,” the latter of which determine the programs or content displayed by a broadcaster’s channels in a certain geographic area at particular times.
In 2022, the Plaintiff sued Fox Broadcasting Company, among other entities, in the District of Delaware for infringement of the four (4) patents. In response, Fox filed a motion to dismiss the complaint based on Title 35 U.S. Code Section 101, arguing that the subject matter of the patents was not within the scope of matter that could be patented (i.e., it was ineligible subject matter) because the subject matter of the patents was an unpatentable “abstract idea.”
The District of Delaware (Judge Williams) granted the motion to dismiss, and the order of dismissal was affirmed on appeal by the Federal Circuit. While the Plaintiff admitted that the patents did not claim a particular machine learning technique, but rather, applied machine learning to a specific context, the Plaintiff’s argument for eligibility was that, “the unique application of machine learning to generate customized algorithms, based on training the machine learning model, that can then be used to create … event schedules that are updated in real-time automatically.” Slip Op. at 8.
Neither the district court nor the appeals court bought this argument. Rather, the Federal Circuit affirmed that a machine learning patent that did not “delineate steps through which the machine learning technology achieves an improvement” (Slip Op. at 13) is not patent eligible. Merely speeding up a human activity (such as developing a network map for a television program in real time) or making a human activity more efficient does not convert otherwise a patent-ineligible “abstract idea” into patent-eligible subject matter.
The takeaway here is that “black box” software inventions (e.g., software patents that do not result in a technological improvement to the computer or computer system, but rather, merely process data or perform a function on a computer rather than manually) are most likely not going to be patent eligible under the current laws in the U.S.
While every invention and every case is different and requires review and analysis by a qualified patent attorney, it is important to consult with qualified intellectual property counsel who has knowledge of the patent system to consider what, if any, protection is available for such inventions through patenting, or to determine what other avenues of intellectual property protection may be available for such inventions.
