The legal landscape for patenting artificial intelligence (AI) and machine learning algorithms has shifted notably in recent years, yet many questions remain unanswered regarding what constitutes a patent-eligible improvement in these areas. This gap was highlighted in the recent decision by the US Court of Appeals for the Federal Circuit in Recentive Analytics v. Fox, which provides insight into the type of claims that may stand a better chance under current law.
A surge in patent applications involving machine learning has been observed, yet legal guidance remains insufficiently clear. The core issue revolves around whether an AI invention can be deemed patentable based purely on improvements within the machine learning model without contributing additional technical advancements beyond the algorithm itself. The lack of clarity presents a formidable challenge for practitioners trying to navigate the current patent framework effectively.
Previous cases, such as Alice Corp. Pty. v. CLS Bank Int’l, set the groundwork by affirming that merely applying a generic computer to an abstract idea doesn’t convert it into a patent-eligible invention. Guidance from the US Patent and Trademark Office (USPTO) reinforced this perspective by outlining scenarios in which machine learning applications would not qualify for patent protection unless they contribute improvements to specific technological sectors.
The Federal Circuit’s stance in Recentive has significant implications for those seeking patent protection in AI and machine learning. The court was unpersuaded by claims that leveraged existing machine learning technologies to boost advertising revenues without showcasing any technical enhancement of the technologies themselves. This judgment illustrates a pivotal point: merely applying machine learning to enhance task efficiency doesn’t meet the threshold established by Alice.
The ruling suggests the necessity of a defined technological improvement within the machine learning model for eligibility, although the specifics of what qualifies as such an improvement remain undefined. The USPTO’s guidance has similarly focused more on connection to broader technological fields rather than on improvements within the algorithm itself.
The current atmosphere leaves practitioners with uncertainty and highlights the need for strategic drafting of patent applications. Successful claims are more likely when they emphasize technical applications or enhancements within traditional technological fields that benefit from machine learning. Practitioners are advised to articulate the technical advancements brought about by their algorithms meticulously and maintain patent family continuity to adapt to legal evolutions.
In summary, while Recentive provides some direction regarding the eligibility of machine learning patents, it stops short of comprehensive clarity, indicating only that a strong connection between the algorithm and broader technical improvements remains pivotal. Practitioners are encouraged to remain agile and consider best practices to dial into these evolving standards.
For further insights, refer to the full article by Lidiya Mishchenko and Pooya Shoghi.