NEWS
USPTO AI subject matter eligibility guidance
The US Patent Office continues to issue guidance on subject matter eligibility under 35 USC 101. This latest July guidance addresses AI-assisted inventions, and adds very little to what we already know. Basically, AI cannot be named as an inventor, and otherwise AI-assisted inventions are prosecuted the same as non-AI assisted inventions.
Invention Subject Matter Eligibility
All kinds of things are eligible for patentability under the guidelines of the USPTO: process, machine, manufacture, or composition of matter. There are some judicial exceptions that the courts have determined are the basic tools of scientific and technological work, and as such, are excluded from patentability since monopolization of these tools would tend to impede innovation more than promote it. Those judicial exceptions are laws of nature, natural phenomena, and abstract ideas.
But the courts have decided that if a patent claim can recite “additional elements” that integrate the judicial exception into a practical application, then such a claim can be eligible for patentability.
First Prong: Claims Directed Towards Judicial Exceptions
In the first prong, there is an evaluation of whether a claim recites an abstract idea.
It is common for claims to AI inventions to involve abstract ideas, but PTO personnel must draw a distinction between a claim that recites an abstract idea, and one that is merely based on an abstract idea.
The USPTO has defined three groupings of abstract ideas: mathematical concepts, methods of organizing human activity, and mental processes.
Mathematical concepts
The mathematical concepts abstract idea grouping is defined as mathematical relationships, mathematical formulas or equations, and mathematical calculations. The claim limitations do not fall within the mathematical concept grouping if it is only based on, or involves, a mathematical concept.
Methods of Organizing Human Activity
The certain methods of organizing human activity abstract idea grouping describes concepts related to fundamental economic principles (including hedging, insurance, mitigating risks); commercial or legal interactions (including agreements in the form of contracts, legal obligations, advertising, marketing or sales activities or behaviors, and business relations); and managing personal behavior or relationships or interactions between people (including social activities, teaching, and following rules or instructions).
Mental Processes
The courts consider a mental process (thinking) that can be performed in the human mind, or by a human using a pen and paper, to be an abstract idea.
Second Prong: Additional Elements
If the claim under evaluation is directed towards one of these judicial exception areas, then the second prong of the evaluation is to identify whether there are any additional elements that would integrate the exception into a practical application. Many claims to AI inventions are eligible as improvements to the functioning of a computer or improvements to another technology or technical field. While the courts haven’t provided an explicit text for how to evaluate the improvements consideration, they have instead illustrated how it is evaluated in numerous decisions.
A key point of distinction for AI inventions is between a claim that reflects an improvement to a computer or other technology (eligible) and a claim in which the additional elements amount to no more than a recitation of the words “apply it” or are no more than instructions to implement a judicial exception to a particular technological field (ineligible).
Eligibility and AI-assisted Inventions
AI assistance is not a consideration in subject matter eligibility analysis. Patent protection can be sought for AI assisted inventions where one or more persons made a significant contribution to the claimed invention. The current guidance does not provide for recognition of tools like AI for inventorship purposes.
The Power of Arbitration Agreements in Patent Royalty Disputes
The recent case, Zimmer Biomet Holding v Insall, resulted in a judgment requiring the licensee to continue paying royalties even after expiration of patent upon which the license agreement was predicated. The holding is clearly in conflict with Supreme Court precedent, but was upheld anyway based upon the arbitrator’s unreviewable interpretation of the underlying agreements.
The Precedents: Brulotte and Kimble
There are two key precedents at issue in the Zimmer dispute: Brulotte v Thys, and Kimble v Marvel.
In Brulotte, the Supreme Court held that it is unlawful per se to have a royalty agreement extending beyond the expiration date of the patent. This was called the Brulotte rule. The rule was widely criticized at the time (1964), and the rule continued to receive criticism through the years. In Kimble (2015), the Court provided a carve out: “post-expiration royalties are allowable so long as tied to a non-patent right”.
The Royalty Agreements
The dispute between Zimmer and Insall centered around a license to a knee replacement technology. The original 1991 agreement defined specifically that Zimmer would pay royalties to Insall on the sale of products covered by Insall’s patents. In 1998, the agreement was updated to include a statement that 1% of royalties would be calculated on sales of Zimmer’s NexGen Knee family of products, not necessarily covered by the patents.
The Case: Zimmer v. Insall
After the Insall patents expired in 2018, Zimmer stopped royalty payments, but continued to sell the products. When challenged, Zimmer argued that Brulotte rendered any post-expiration royalties unenforceable.
The parties went to arbitration in 2019, and the arbitration panel sided with Insall. The amendments to the royalty agreement effectively divorced royalties from Insall’s patents, and instead married them to NexGen marketing.
The issue was brought to the district court when Zimmer tried to get the arbitration award vacated on the basis that it violated the public policy established in Brulotte. The Insall estate motioned to dismiss and asked the court to uphold the arbitration’s decision. The district court agreed with the Insall estate and confirmed the arbitration award.
Zimmer then appealed this to the Seventh Circuit court. The Circuit court stressed that “arbitration awards are largely immune from…scrutiny in court” under the Federal Arbitration Act. Under that act, arbitration awards can generally only be appealed on very narrow grounds, including a public policy exception where the award violates some well-defined and dominant, explicit public policy, an arbitral award can be vacated. And even then, a court is bound by the arbitrator’s interpretation of the underlying contract in such an evaluation.
The Consequence
The essential question was what the royalties were based on. If they were based on Insall’s patent rights, as the original 1991 royalty agreement stipulated, then the agreement was unenforceable under Brulotte. If they were based on non-patent rights, as the 1998 amendment stipulated, they would be permissible under Kimble.
The Circuit court decided the 1998 amendments were within the arbitrator’s unreviewable authority.
Zimmer tried arguing that the amendment changed how the royalties were calculated, not why they were paid in the first place. But the court noted an “arbitration clause delegates interpretive power to the arbitrators. We don’t ask whether they read the language correctly, it is enough that they tried to apply the contract that the parties signed.”
Amazon APEX and Personal Jurisdiction
In May 2024, the Federal Circuit issued a decision concerning a patentee’s use of Amazon’s patent enforcement process, called APEX.
Briefly, Amazon has a low-cost procedure, the Amazon Patent Evaluation Express Procedure, or APEX, that is meant to resolve patent infringement issues for products sold on Amazon’s website. The process is meant to take only 7 weeks to get a final disposition, and costs $4000. There are several other private sales-channel enforcement processes that are becoming more popular, given the low cost and relative quickness of getting a resolution.
LDG, a Delaware LLC with headquarters in Arizona, is the owner of a patented electrical receptacle with additional functions. SnapPower, a Utah company designs, markets and sells electrical outlets as well, and uses Amazon.com as a sales channel.
In May 2022, LDG submitted an APEX challenge alleging SnapPower sold products on Amazon infringing one of LDG’s patents. Amazon notified SnapPower and both LDG and SnapPower exchanged emails and held a conference call but came to no agreements.
SnapPower subsequently filed an action for declaratory judgment of non-infringement in its home state of Utah. LDG asked to have this dismissed for lack of personal jurisdiction, which the district court granted. The court said LDG’s allegations were directed towards Amazon in Washington, where the APEX agreement was sent.
But the Federal Circuit reversed the district court’s decision on jurisdiction.
The court laid out a three-factor test for whether specific personal jurisdiction comports with due process:
1) whether the defendant purposefully directed its activities at residents in the forum,
2) whether the claim arises out of, or relates to, the defendants’ activities in the forum, and
3) whether assertion of personal jurisdiction is reasonable and fair.
The first two factors comprise the ‘minimum contacts’ portion of the jurisdictional framework. If those are satisfied, then specific jurisdiction is “presumptively reasonable”. The burden then shifts to the defendant to present a compelling case that the presence of some other considerations would render jurisdiction unreasonable.
SnapPower argues that LDG purposefully directed enforcement activities at Utah when it initiated the APEX program. The court agrees that LDG did so, intending effects which would be felt in Utah, so the first element is satisfied.
LDG argues SnapPower’s action for declaratory judgment of non-infringement does not arise from, or relate to, any activity by LDG in Utah because the APEX agreement was sent to Washington, not Utah.
The court held however, that LDG’s actions in submitting the APEX agreement was directed towards SnapPower in Utah and aimed to affect marketing, sales, and other activities in Utah, SnapPower’s suit arises out of the defendants’ activities within the forum. The second element is now satisfied.
With the first two elements being satisfied, specific jurisdiction is presumed reasonable.
LDG, however, argues it is unfair and unreasonable. Ruling for SnapPower in this instance would open the floodgates of personal jurisdiction and allow suits against any APEX participant anywhere in the country. The district court agreed with LDG, noting that principles of fair play afford a patentee latitude of its patent rights without subjecting itself to jurisdiction in a foreign forum. The Federal Circuit, however, disagreed and ruled that LDG did not meet its burden to present a compelling case of other considerations that would render jurisdiction unreasonable. First, the Federal Circuit contested that the floodgates would be opened on personal jurisdiction, and second, disagreed that such a thing would be inherently unreasonable. The Federal Circuit said LDG initiated a process in APEX that, if SnapPower had taken no action, would have affected sales in Utah. LDG has not articulated a compelling argument for why it would be unfair or unreasonable for it to be subject to specific personal jurisdiction in Utah under these circumstances.
Cellect v. Vidal and the future of Patent Term Adjustment
The U.S. Supreme Court is currently considering Cellect v. Vidal, a case with potential impact on many patents.
Cellect is the owner of 4 patents being challenged, each of which is directed towards devices comprising image sensors. The patents are all part of one patent family claiming priority from a single patent application. Because of delays during prosecution, each of the challenged patents was granted patent term adjustments (PTA) of varying lengths.
Cellect sued Samsung for infringement of these patents. Samsung responded by seeking ex-parte reexaminations, asserting that, due to Obviousness-Type Double Patenting (ODP), the patents were invalid. During the reexamination, the examiner determined that the claims were obvious variants of the prior expiring reference patent claims, and the claims were invalidated.
Cellect appealed the rejection, arguing that ODP doesn’t invalidate a validly obtained patent term extension, and therefore shouldn’t negate a statutory grant of patent term adjustment.
Some background in case the reader isn’t a patent attorney.
Prior to 1995, a patent term was 17 years from the date of issuance. Accordingly, delays at the Patent and Trademark Office (USPTO) didn’t impact patent terms because those terms began at issuance.
In 1995 the United States passed the Uruguay Round Agreements Act, which set patent terms to 20 years from earliest claimed priority date. Since all patents in a family have the same earliest priority date, it follows they would have the same expiration date. This meant that after the URAA, delays at the USPTO meant cutting into the effective term of a patent.
In 1999, the Patent Term Guarantee act, part of the American Inventors Protection Act, promised inventors a patent term adjustment period for examination delays caused by the PTO. Given the amount of delays, the PTO adjusted approximately half of the patents issued between 2005 and 2022.
One of the effects of this has been that patents in the same family routinely expire later than others due to different delays in application prosecution, and their mandated adjustments.
So what’s the problem?
There is a ‘double patenting’ doctrine that bars applicants from receiving more than one patent on the same invention. Way back, inventors could theoretically extend their exclusivity over an invention by claiming obvious variations of the same thing. Staggering these applications could give the inventor a longer monopoly than was justified under the principles of the patent system. To rectify that problem the Courts developed a doctrine called obviousness-type double patenting (OPD). That doctrine was designed to prevent unjustified extensions where a later patent is merely an obvious version of a prior patent in the family. It is the extension of patent terms that is the sole basis for the doctrine.
However, if an applicant has two patents on indistinct inventions, the applicant can file a terminal disclaimer to make the patents expire on the same day. If the patent owner files the terminal disclaimer, then no ODP. But after the first patent expires, ODP invalidates the patent expiring later.
In 1984, Congress authorized patent term extensions (PTE) to compensate pharmaceutical patent holders when the FDA approval process would bar the invention from the commercial market. PTE would extend the patent term for the amount of time lost during that approval period. This is a different mechanism than PTA, which adjusts patent terms due to USPTO delays.
Back to the situation as it stands.
Where does this leave ODP? The point of ODP was to stop gaming of the system to extend patent terms. But both PTA and PTE are congressionally authorized compensations for time lost due to outside factors, not attempts to game the system. The Federal Circuit affirmed in 2007 and 2018 that differences in expiration dates caused by PTE did not trigger ODP. But the relation of ODP to PTA has been less certain.
Cellect challenged the decision before the Federal Circuit, arguing that the Board had wrongly expanded the scope of ODP, and punished patent owners whose terms had been adjusted. But the Federal Circuit upheld the decision.
The case is now before the Supreme Court and the stakes are high. The decision as it stands would largely undo Congress’s Patent Term Guarantee Act. Given the amount of patent terms that have been adjusted, there are a lot of patents that could be invalidated solely on the bases of those adjustments.
PTO Guidance on AI and Inventorship
Questions about who, or what, can claim inventorship on AI-assisted inventions have been tested in the last few years. The USPTO has issued a 27-page pdf to give guidance on the question of “how inventorship should be determined in U.S. patents and patent applications when AI, including generative AI, plays a role in the conception and/or reduction to practice of a given invention.”
The short version is that nothing has changed- only natural persons are considered as possible inventors.
That goes for, “joint inventors” or “coinventors” as well. Attempting to list a machine on an Application Data Sheet will be seen as improper inventorship.
Back in 2019, the Patent Office crowdsourced the public’s thoughts on AI-assisted inventions. It published a report on the findings in 2020, and in 2022 and 2023, the PTO held conferences where the major issues were considered.
A recent Executive Order was sent out recognizing the great potential of AI to solve some difficult problems but also calling for guidelines to both encourage a competitive ecosystem and guard against emerging risks. The USPTO guidance is a direct response to that Executive Order.
The USPTO, while affirming that inventors must be natural persons, does acknowledge that “AI systems could perform acts that, if performed by a human, could constitute inventorship under existing U.S. patent law”. It also states that “AI-assisted are not categorically unpatentable, but inventorship analysis should focus on human contributions as patents function to incentivize and reward human ingenuity”.
The stress is that the U.S. patent system is designed to encourage human ingenuity.
Some of the guiding principles in determining AI-assisted inventorship include:
- A natural person’s use of an AI system in creating an AI-assisted invention doesn’t negate the person’s contribution as an inventor.
- Merely recognizing a problem doesn’t rise to the level of conception. A natural person who only presents a problem to an AI system may not be a proper inventor.
- Reducing an invention to practice alone is not a significant contribution that rises to the level of inventorship.
- A natural person who designs, builds, or trains an AI system in view of a specific problem to elicit a particular solution could be an inventor, if that person’s contribution was significant.
- Maintaining intellectual domination over an AI system does not, in itself, make a person an inventor of any inventions created through the use of AI. Simply owning the system doesn’t make the person an inventor.
The guidance also stresses the duty of being forthcoming about the contributions that humans made in an AI-assisted inventions.
The discussion in the guidance clearly references Stephen Thaler and his attempts to get his DABUS system recognized as an inventor. The issues at stake for the PTO are that the patent system is conceived as an incentive and encouragement for human ingenuity and stimulus to sharing of knowledge. Including an AI system as an inventor isn’t going to stimulate more ingenuity in the AI system, so there is no point in it.
Nor is an AI system going to seek protection from infringement on any hypothetical patents to its name.
The rest of the guidelines are trying to find the balance between the input of human and machine in the process of innovation.
The full guidance can be found here: https://public-inspection.federalregister.gov/2024-02623.pdf
Speeding Up Patent Prosecution
We have uploaded our latest FishFAQ™ video to YouTube. This video covers various options for speeding up patent prosecution. You can watch it on a small scale here or follow the link on the video face to see it on YouTube.
Creativity: Human and AI
Dr. Stephen Thaler is basically challenging the need for the ‘human’ restriction in the law. In the recent case of Thaler v Perlmutter, he requested reconsideration of the Copyright Office’s rejection because, he said, while the Creativity Machine output lacked human authorship it ‘otherwise meets authorship criteria.”
What is the essential difference between the way an AI platform generates a work, and the way a human generates a work? WHY does the law consider only human output eligible for intellectual property rights protection?
AI output generation
An AI system analyzes a ton of data, then looks for patterns. It can then replicate those patterns according to specific prompts. For example, if I ask an AI system to make a painting of a cat playing with string in the style of Van Gogh, the AI can produce such a painting in a relatively convincing style.
Human output generation
On one level, humans do the same thing. My experiences in the world are the data I have to work with. For example, if I want to paint a portrait, I have seen enough faces to know what a face looks like. I can create a face picture from knowing the basic patterns.
Human data is multi-sensual
But human creativity and imagination are closely related to emotions and subjective experiences.
Humans experience things through multiple senses. I’ve not only seen an angry face, I’ve experienced what it means to have a person angry at me. I’ve personally felt my emotional response to that anger, so when I make a picture of an angry face, there is much more than just the shape of the facial features that informs my understanding of what is happening, and therefore what it should look like.
How exactly will that make the painting of the face LOOK any different you might ask… well that will depend on the artist.
Take another example: If I make a picture of the view from my office on the 8th floor, that view might be a straightforward landscape… or it might portray some of the awe of being up so high… or it might portray some of the fear of falling from such a height.
Humans experience things firsthand through all their senses, which gives them a broader range of experiential data with which to work and provides depth unavailable to AI.
Human data is limited and therefore more focused
The very thing that gives AI certain advantages, an enormous set of data to work with and the processing power to sort through it all, may also, paradoxically enough, give humans their creative edge: humans are working with a much narrower set of data. That limitation will provide a more subjective and particular focus, which will provide a stylistic advantage. Of course, specific prompts can be given to the AI to progressively narrow the focus, but those human-introduced prompts are the intrusion of the human into the process.
Creativity and Imagination
Creativity is defined as relating to or involving the imagination or original ideas, especially in the production of an artistic work.
What is imagination?
The faculty or action of forming new ideas, or images or concepts of external objects not present to the senses.
This concept- forming new ideas not present to the senses adds another layer of depth to it. Could AI look at a billion photos of anxious humans and ever come up with Edvard Munch’s The Scream?
What humans do is more than just take input from multiple senses, which maybe increased computational power and more in-depth algorithms could possibly mimic; it is using the personal, sensual experience to express something not present to the senses.