This post is part of our Ask Me Anything series “IP Law for Femtech Founders” with Marshall, Gerstein & Borun Giordana Mahn.
What are strategies for femtech startups looking to protect products based on algorithms or machine learning approaches? Are patents feasible?
This is, actually, rather a complex question– generally, machine-learning systems and methods are patent-eligible, meaning patents are feasible in this area, but this falls under an area of patent law that is in flux, at least in the US, and is subject to change with the advancement of AI.
To start with the patent basics, a patent in the US gives its owner the right to exclude others from making, using, selling, and importing the patented subject matter on US territory. Owning a US patent doesn’t automatically give you the right to do any of those things (i.e., making, using, selling, importing), but just the right to exclude others from doing so! A US patent also doesn’t give you the right to exclude others from making your invention in other countries unless you have patent protection in those countries, as US patents only extend to the US and its territories.
In the US, you can apply for a patent on patent eligible subject matter, which includes processes, machines, manufactures, and compositions of matter. Generally, something is patent eligible if it is a method or a thing, and not an idea. However, a patent application will only be granted if the US Patent Office deems that your invention (a) doesn’t fall into one of the exceptions to patent eligibility (listed below), (b) is new, and (c) is not obvious in view of what’s in the public space.
Exceptions to patent eligibility, in the US are (1) abstract ideas, (2) laws of nature, and (3) natural phenomena. US courts have decided that subject matter that fall under these categories are considered basic tools for innovation, and therefore shouldn’t be subject to a monopoly. Algorithms or mathematical concepts fall under these categories. Think about it, if someone received a patent on a law of physics, they could exclude you and everyone else from using it to develop technology. That wouldn’t be fair, and would have detrimental effects to innovation. So the US patent office and courts take a hard look at patent claims that recite or include algorithms to make sure it doesn’t cover one of these basic tools.
But that doesn’t mean you can get patent protection over products based on algorithms or machine-learning. Depending on the type of technology, there are many different strategies for claiming your invention so that it doesn’t fall under one of these exceptions to patentability. One example is to provide a practical application of that algorithm in a way that is “something more” than just a computer performing an algorithm. If any of this is at all confusing, that is because this is a confusing area of law, and the US courts and patent practitioners have wrestled with these issues for years.
I also want to mention the exceptions to patentability in Europe, as I know Femtech Insider subscribers are tuning in from across the Atlantic. Those exceptions are (1) inventions that are contrary to public morality, (2) plant or animal varieties or biological processes for the production of plants and animals, and (3) methods for treatment of the human or animal body by surgery or therapy and diagnostic methods practiced on the human or animal body. This third exception to patentability is a major difference between US and European patent eligibility requirements. See https://www.epo.org/law-practice/legal-texts/html/epc/2016/e/ar53.html (although the US also has some limited exemptions to liability for patent infringement for medical practitioners performing medical activities, 35 USC § 287(c)).
As for machine learning systems and methods, the USPTO provides guidelines at their website. At the following link, the USPTO provides updated guidance related to a machine learning method claim. https://www.uspto.gov/sites/default/files/documents/101_examples_37to42_20190107.pdf (see Example 39).
New questions in patent law are popping up as it relates to AI – for example, can a machine be an inventor? While the USPTO has said “no”, machines might start inventing sooner than we think.
DISCLAIMER: These comments do not necessarily reflect the views of Giordana Mahn’s firm of Marshall, Gerstein & Borun or its clients, and should not be relied upon as legal advice for any particular situations. Providing these responses is not intended to and should not be considered as reflecting an attorney-client relationship.