INNOVATE Magazine
INNOVATE is the online magazine by and for AIPLA members from IP law students all the way through retired practitioners. Designed as an online publication, INNOVATE features magazine-like articles on a wide variety of topics in IP law.
The views and opinions expressed in these articles are those of the authors and do not necessarily reflect the views or positions of AIPLA.
Articles
In This Section
Where Are All the Software Output Lawsuits?
Evan Zimmerman
Intellectual property rights have a general requirement that they come from people. In patent law, this is called inventorship. In copyright law, authorship. Although these rights can be assigned to corporations, in general the position of nearly every government office is converging on the opinion that inventors and authors must be people. The stakes are particularly high because failing to include an individual inventor or author means the death of that IP.
In the patent world, the USPTO recently put forward guidance on the matter. Under the inventorship requirements set out under §§ 35 U.S.C. 115-16, the Patent Office drew on Pannu v. Iolab Corp. 155 F.3d 1344 (Fed. Cir. 1998) to set out a “significant contribution” test. In essence, to satisfy the bar of a natural inventor, each inventor must have (1) “contributed in some significant manner” to the conception or reduction to practice; (2) made a contribution that is “not insignificant in quality” measured against the “dimension of the full invention”; and (3) do more than just explaining the state of the art to the real inventors. Under Thaler v. Vidal 43 F.4th 1207 (Fed. Cir. 2022), an AI cannot patent an invention. Therefore, under these rules, it is possible to have a patent application with no inventor deemed invalid.
Yet, tech commentators, like Marc Andreessen, argue that AI is just software. And you are allowed to use software when creating an invention. So it would stand to reason that you should be able to use AI when creating inventions. This distinction, therefore, leads to a quandary.
This raises the question whether there are equivalent lawsuits for other types of intellectual property generated by software that could help illuminate this given that there is now an AI inventorship case, Thaler v. Vidal.
After all, as a general matter, software is essential for the development of new intellectual property. In the pharmaceutical industry, for example, it is common to use drug screening algorithms, some of which are powered by machine learning, to determine which drug candidates are worthy of further study. One such drug, halicin, was developed at MIT after the use of an algorithm that screened hundreds of millions of candidates in a few days. But it’s not just pharma. Aerospace simulation, for example, is central to the development of new aircraft. The Boom XB-1, the world’s first new supersonic commercial plane, makes extensive use of simulation. Even in copyright, lawyers Damien Riehl and Noah Rubin published, for fair use purposes, 68 billion music files supposedly containing all possible melodies.
Though inventorship law is obscure, it does come up, whether in recent cases like HIP, Inc. v. Hormel Foods 66 F. 4th 1346 (Fed. Cir. 2023) or older ones like Stanford University v. Roche Molecular Systems, Inc. 563 U.S. 776 (2011). And so one may have expected some enterprising lawyer, somewhere, to challenge software-based inventorship in a high-stakes industry. Certainly, that issue is worth discussing now.
In addressing this issue, I had hoped to find a precedential lawsuit involving computer-assisted design or the results of a simulation. Perhaps, someone had argued that an algorithm should have been an inventor, or that a simulation meant that an invention was “obvious to try,” per KSR v. Teleflex 550 U.S. 398 (2007).
However, these cases simply don’t exist in patent law. No one has ever litigated this inventorship issue involving computer-assisted design in any jurisdiction. In fact, I have previously found that there have been more law review articles on the topic of the impact of AI on inventorship than on software generally. The reason is relatively simple. When it comes to regular software, the idea that software is the inventor is so absurd that no one has raised it.
I was able to find a single case in the U.S. on the matter in the case of copyright, but it is wholly unedifying. In Design Data Corp. v. Unigate Enterprise, No. 14-16701 (9th Cir. 2017) Design Data Corporation (“DDC”) claimed that Unigate Enterprise (“UE”) infringed on its copyright regarding its CAD software, claiming that it pirated its software for use with certain clients. Interestingly, however, DDC argued that not only was its software pirated: so was each of the outputs due to the unique data files that DDC’s software generated. In other words, DDC claimed that the outputs were also copyrighted, and so UE was liable for infringement not only for copying its software but for the generation of DDC’s proprietary file format every time UE used the infringing software.
The Ninth Circuit affirmed the lower court’s ruling that that the software, but not the outputs, were protectable. Furthermore, in Design Data Corp., the Ninth Circuit suggested, in dicta, that while this finding was fact-specific, there could be a broader precedential ruling in the future. The Court indicated that in the future it might find that protection may run with the program if it “does the lion’s share of the work” and the user’s role is “so ‘marginal’ that the output reflects the program’s contents.” This is intriguing because it suggests that, at least in copyright, sufficiently advanced software may have its output protected by copyright on behalf of the inventor of the software. This is the opposite of what we might expect given the notices from various global Copyright Offices, like in the United States, that AI, as a matter of law, cannot constitute an author. Yet, to date, Design Data remains an isolated, judicial curiosity.
For patent law, the only case on point was one obscure case in Europe that failed to clarify this inventorship issue. Specifically, Decision G 0001/19 (Pedestrian simulation) 10-03-2021 found that a simulation is not patentable in and of itself, but that is not the same as asking whether the output of a simulation is patentable. In fact, the Court held that a simulation’s patentability was unaffected by whether it was part of a design process. Although this case was an attempt to clarify a law regarding the patentability of simulations, it left open the problem of what is a simulation. Are large language models not, in some sense, merely probabilistic simulations of embeddings? Where is the line—can CAD software, if it does enough of the work, count as a “simulation?”
Unfortunately, the existing guidance risks generating legal rules that could disrupt the use of software widely used in inventing today. A natural person who “merely recognizes” the output of an AI system may not be deemed an inventor. But what about machine learning researchers who develop screening algorithms? A natural person who develops an “essential building block” may qualify (but only sometimes; the USPTO does not clarify what those circumstances are) but how essential are those building blocks, especially since modern software development is so dependent on building on the work of others?
Furthermore, how would this be policed? The USPTO says that it is not going to change its rules regarding duty of candor, but with Thaler, this will surely be a topic of litigation going forward. One of the benefits of the AIA was that lab journals and record-keeping were no longer as important for determining priority. But there is a possibility that this approach could require detailed records as to how an invention was conceived in order to fend off an inventorship challenge. Suddenly, the lab journal could be back with a vengeance.
One radical idea would be to allow AI systems to be listed as inventors, but this would not solve the core line-drawing problem. Yes, this would make the issue of policing easier because the incentive to disclose would be higher. But there would still be numerous topics of debate, such as: What is the difference between AI software and a mere simulation? Do inventors need to disclose non-AI software aids? If they don’t, did they fail to properly disclose an inventor and thus forfeit patentability? If AI can be listed as an inventor, does that mean that non-AI software is suspect? The core line to be drawn, it seems, is not whether AI must be listed as an inventor but rather what counts as AI, and that is not where anyone from courts to patent offices have been willing to tread. Although, to be fair, even the definition of AI is in flux.
This seems like a solution in search of a problem. Thus far, the only machine-learning/AI systems that are capable of producing new inventions are those that are highly customized and designed by inventors for their scientific or engineering purposes, which are precisely the types of specialized software that have been used for decades with no concerns raised. When this type of general-purpose AI is developed there will be bigger questions, both in the patent world and the world at large, than whether that AI system should count as an inventor.
Evan J. Zimmerman is an entrepreneur, investor, and writer. Evan is the co-founder of Edge, a patent assistant software company that leverages AI, for which Evan was recognized as a Forbes 30 Under 30 lister in 2023 and backed by Y Combinator. Prior to that, Evan was Chairman of the Clinton Health Access Initiative technology council, which advises the technology of global public health in dozens of partner countries and is a member of the strategy board for Broad Center for Regenerative Medicine at USC. He is also the founder of Jovono, a venture capital firm, and was also a cofounder of a consumer products company that licensed patents in multiple successful areas. Evan also speaks and writes on technology, with publications in places like Techcrunch, Stat News, Palladium Magazine, the California Management Review, the South China Morning Post, and others. He was inducted as the youngest member of the MAK Museum in Vienna's Biennale Circle, which planned the 2017 Vienna Biennale on the “Future Of Work. Evan has a law degree from Berkeley Law School, where he was a Dean’s Scholar with certificates in technology law and business law and won a Prosser award. He graduated from the University of Chicago, where he was a University Scholar, with general and departmental honors in economics and conducted his thesis work on the economics of embargoes as a visiting undergraduate at Harvard University.
Innovate Volume 18 Timeline
submit articles to innovate@aipla.org
Submission Window Open
January 3, 2025
Submission Deadline
April 11, 2025
Publication Date
June 13, 2025
About
Publishing an article to INNOVATE is a great way for AIPLA members to build their brand by increasing recognition among peers and setting themselves apart as thought leaders in the IP industry.
Any current AIPLA member in good standing may submit an article for consideration in INNOVATE throughout the year. IP law students are especially encouraged to submit articles for publication.
Articles submitted to innovate@aipla.org are reviewed by an ad-hoc sub-committee of volunteers from AIPLA's Fellows Committee, and other AIPLA peers.
Don’t miss your chance to be published with AIPLA’s INNOVATE! Email your article submission to innovate@aipla.org to be considered for the next edition.
For more information please review the Guidelines for Article Submission and the INNOVATE Author Acknowledgement Letter for guidelines and terms of article submission and publication.