Edge now solve this problem with formalize.

When you go to the Figures tab, every type of figure generation and upload is now unified.

You can now upload figures or create illustrations from scratch. This includes flow charts, text-to-image for in-use figures, and our standard formalization. Edge can also formalize line drawings and charts. No more hatching required.

When you upload an image, Edge can detect on its own whether an image is a chart or a drawing. The only time you might want to manually select it is when you have a chart that you want represented literally, such as a design for a user interface that includes a chart.

Then just wait—it usually takes 20 seconds for line drawings and 45 seconds for charts—and you have a submittable drawing.

You can also edit drawings. You can both prompt the Assistant and circle portions of the drawing.

The first post in this series is about the rise of the MSO, or Managed Service Organization. What is an MSO? It is an operating company owned at least in part by non-professional capital. It purchases equipment and software in bulk, negotiates leases, employs administrative staff, and more. Because it is technically not a professional practice, it can also raise equity capital and is typically easier to underwrite for debt by creditors. In exchange for all it does, although the MSO typically does not directly own shares in the professional practice, it extracts fees—often 90%+—that equal de facto ownership.

Lawyers are just starting to hear about them. What are they, and why are they emerging now? In this post, I argue that the future law firm will increasingly use MSOs and that there is much to learn from the medical field, which pioneered the use of MSOs in a profession.

The Rise of Medical Capital

Historically, physicians were small business owners. But now things are increasingly consolidating. At the same time, the medical profession has boomed, and medical spending is now nearly 20% of GDP. Why? And are these related?

The story you’ll hear in the press is about private equity driving up prices and pushing consolidation, but I believe the story of this capital flow is typically told backwards, however, confusing cause and effect. Usually, the story is that the rules got loosened, capital flowed in, and this changed the medical profession. But I believe it was inevitable because of the changes in the practice of medicine. To understand why, let’s look at the history of medicine in the United States.

After World War II, the United States embarked on a journey of heavy scientific research, especially in medicine, with the institution of the NIH and NSF, alongside specific efforts like President Richard Nixon’s war on cancer. The result was the rise of specialty care and increased capital costs as medicine got more sophisticated. You were naturally going to require more capital to fund this specialization and technology.

Still, the Normal Rockwell image of the home doctor is what most people would have experienced before our medical revolution. This changed rapidly, and you can see it in the data.

Even after the standardization of the medical industry after the transformational Flexner Report in 1910, in 1931 84% of doctors were still general practitioners; by 1965, only 37% were. In 1931, there were 17 specialties; today, there are 40 specialties and 89 subspecialties. This brought up the capital costs needed to start a practice as well as the operational complexity needed to practice medicine.

With specialization, suddenly it wasn’t enough to have a community, an office, and a stethoscope to practice medicine. You needed to work with complex insurance schemes, you needed more specialized space, and you needed to spend possibly millions of dollars on specialized equipment just to practice your area of medicine.

Increasing complexity is usually dealt with through economies of scale. Larger organizations are able to spread out operational costs and complexities through a larger revenue base. And so it was in medicine: increased specialization drove consolidation.

The problem is that while medical training in the US is the best in the world, meaning that most doctors are good at their jobs, they are often not businesspeople. That means that in our hyper digital world, they are bad at marketing themselves, bad at business operations, and just often bad at the things you need to do to run a business.

This naturally attracts capital to the medical business, as it would any business field.

While what you may have in mind is the growth of private enterprises, like HCA, and giant hospital systems that seem larger than ever (note how many states have a hospital as its largest employer), the other main effect has been physicians buying each others’ practices. Taking on capital partners, business-savvy doctors are buying other practices 5-10 at a time and creating mini local empires of practices. This has also allowed them to bring in operating and technology partners.

The problem is that all of this is very, very expensive. Thus was the rise of the MSO: it was a response not to scale but to a need for capital.

While a few states, like New York and New Jersey, are still relatively restrictive on practice ownership, most states like California only require majority 51% ownership by physicians. Some states, like Florida and Illinois, have no such restrictions at all. And everywhere, this restriction is skirted through MSOs. This change began in 1982 as a result of the famous case American Medical Assn. v. FTC, 455 U.S. 676 (1982), where the FTC won a victory against the AMA for restricting non-physicians from any involvement in their practice.1 And while fee sharing is heavily regulated under the Anti-Kickback Statute and Stark Law, fee splitting is looser in medicine than in the law.2

Today private equity owns 30% market share in at least one specialty in 31% of all US metropolitan areas. This arrangement all works through MSOs.

What’s Next for Law Firms?

You’re already seeing things change. Arizona and Utah were the first. Soon Texas will be next. The dam is about to break on so-called sandboxes.

This sounds like the story we saw before with medicine, but there is a very key difference. Lawyers are extremely strict about sharing the money with nonlawyers, even aside from kickback concerns. Ethics rules prevent any fee sharing with non-lawyers, even referral fees, under ABA Model Rule 5.4, which most jurisdictions follow. Similarly, they do not allow for even a quantum of non-lawyer ownership under independence rules. These rules have restricted access to capital and business model innovation. In contrast, medicine has been able to allow massive flows of much-needed capital to enter the healthcare industry as a result of better-configured rules.

AI is going to change the economics of the law practice. It will drive profitability and increase revenue, but it will also change some of the required skills and training practices, increase the operating costs of running a law firm, and increase the returns to business savviness. Think of it as raising the barrier to entry, increasing the returns to scale and modernization, and ultimately rewarding law firms that are better run from a business perspective.

The legal profession underwent a change in specialization during the same period as the medical profession did. I would actually go further and say that the legal field is 100% specialized today; I wouldn’t even know what to call a “general” lawyer. There are IP lawyers, personal injury lawyers, employment lawyers, tax lawyers, criminal defense lawyers, trust and estate lawyers, securities lawyers, and much more. Even “corporate lawyers” only handle a subset of corporate law.

And yet firms have gotten much larger without requiring outside capital. Why? The reason is that the law is abstract so increasing specialization did not require increased operational complexity or capital requirements. A great litigation partner does not require a more expensive Lexis Nexis subscription than the next person, while a neurosurgeon will require ten million dollars of equipment to even start doing their work that is different than the ten million dollars of equipment required by an oncologist. Although the value of brand and economies of scale for administrative tasks justifies consolidation, it does not require outside capital in and of itself, and thus no MSO was necessary, and the legal profession remains more fragmented than the medical profession. With AI, this is no longer true. Extraordinarily expensive software, specialized per practice, is becoming the norm. AI is the equivalent of medical equipment for lawyers.

This is going to drive much more M&A, as we’ve already seen in the news. But it won’t just be big firms. Many larger boutiques are going to want to merge—who wants to be two 50-lawyer firms with a narrow focus when you can be one 100-lawyer firm that is full-service in your niche? MSOs will be a key part of this, not only because they will make it easier for firms to partner as collectives but more importantly because it will allow law firms to access capital and talent. Take Crosby, a new-age law firm. Because they have an MSO, they can offer stock options to the best technical talent in the world and raise venture capital, which they did from Sequoia, one of the world’s leading venture capital firms. Notably, McDermott Will & Schulte LLP, number 23 of the Am Law, is now considering an MSO specifically for the purpose of bringing in private equity—after all, they just bought Schulte Roth & Zabel LLP. Neither of these firms will be the last.

I do believe, as well, that this will also allow for an increase in collaboration between law firms and technology companies. Why is there no ZipRecruiter for law firms? Because ZipRecruiter could not charge a fee, which destroys the market for both sides—that’s fee splitting for the lawyers and the elimination of the incentive for ZipRecruiter. I believe that many technology companies are forced into an unnatural arrangement where they are forced to try and sell software to lawyers due to this rule. There are many businesses that would be better off as a marketplace charging a fee, or charging tied to completed work, or as a top-of-funnel referral network. All of these would fall afoul of Rule 5.4 but are acceptable with MSOs. In other words, the MSO should also allow for more experimentation in business models, which the law sorely needs.

Conclusion

So where does this leave law firms?

The legal industry is currently where medicine was in the 1970s: highly fragmented, resistant to outside capital, and clinging to a “craftsman” identity even as it has scaled significantly. But as the “scientific revolution” of AI raises the cost of entry and the complexity of operations, the lone practitioner will become as rare as the 1930s GP.

The future of law isn’t just “more computers.” It’s a fundamental restructuring of who owns the firm, who shares the profits, and how legal expertise is delivered at scale. It will leave the legal professional more efficient and better run.

Most of these suits are seeing their claims dismissed or settled, so there has not been much opportunity to test these legal theories in court. The first of these came down in February, Thomson Reuters Enter. Ctr. GmbH et al. v. ROSS Intelligence Inc., No. 1:20-cv-00613-SB, 2025 WL 458520 (D. Del. Feb. 11, 2025). This week, we got two decisions which may turn out to be seminal: Bartz v. Anthropic PBC, 3:24-cv-05417, (N.D. Cal. Jun. 23, 2025) and Kadrey v. Meta Platforms, Inc., 3:23-cv-03417, (N.D. Cal. Jun. 25, 2025). This week on Nonobvious, we will review these cases and what they mean from the perspective of fair use and AI.

Where We’re Going, We Don’t Need Unfair Use

Without the ability to train a model, there is no AI. Therefore, if rightsholders can prevent model providers from training on their data, they can prevent the creation of AI. Unlike a library, where you can remove the offending book, models “embed” data in a nonliteral way as part of a broader network of understanding; similar to how a human cannot unlearn something they’ve read, a model training run, which costs tens of millions of dollars, cannot be redone to remove the knowledge. The upshot is that a finding against fair use would result in an unworkable remedy or allow a rightsholder to remove their corpus from the body training, which would affect the quality of all models going forward. Add to the complexity that the scale of the corpus is more important than any particular work and this all gets complex fast.

As a reminder, there are four factors: (1) the character of the use, with a focus on whether the work is transformative and commercial; (2) the nature of the copyrighted work, with a balance in favor of fair use in factual contexts; (3) the amount amount and substantiality of the portion copied; and (4) the effect of the use on the potential market. While factor four is typically the most important, the others do still matter, especially the transformative nature of the work.2

AI is a particularly interesting case for fair use. It involves making literal copies of works and potentially transforming them into digitally useful forms; the works involve not just facts but also the most creative works, like poetry and music; there is debate on how transformative the models are, particularly since they can be tricked into “regurgitating” portions of complete works; and because of the potential impact on the market for the works themselves. Adding to the matter is that modern foundational models require truly gargantuan amounts of data, which is rumored to involve the use of well-known datasets with entries of, let’s say, sometimes dubious provenance.

Factually, the way the data was obtained is the main difference between the Anthropic and Meta cases. In the case of Anthropic, the company used a combination of admittedly pirated works obtained in a digitally native fashion—in this case, the subtly named Pirate Library Mirror—as well as purchasing over two million used books which it then destructively digitized. In the case of Meta, famous authors including Sarah Silverman objected to the mere inclusion of their book in the dataset, regardless of how it was obtained (though in the case of Meta, although it is alleged that they use the notorious pirated library LibGen, there was no discovery to prove it). Otherwise, the question of fair use is basically the same.

Digitization was only an issue in Anthropic, and Judge Alsup held those efforts to be allowed as fair use under precedents like Sony Computer Entertainment, Inc. v. Connectix Corp., 203 F.3d 596 (9th Cir. 2000) and Sega Enterprises Ltd. v. Accolade,
Inc., 977 F.2d 1510 (9th Cir. 1992), which focus on the “ultimate use” of the copying. Judge Alsup, in particular, focused on the lack of the “multiplication” problem present in A&M Records, Inc. v. Napster, Inc., 239 F.3d 1004 (9th Cir. 2001) because Napster distributed sometimes millions of copies of the works while Anthropic was merely changing the form of the works for its own use. When it comes to piracy, however, Judge Alsup clearly comes down hard against it. In Meta Platforms, Judge Chhabria surprisingly does not hold Meta liable for downloading “shadow” copies of works they have purchased separately.3

Both cases grant partial motions for summary judgement, finding that in these cases, training a large language model on a corpus is fair use—notably, this case is not about the outputs of the models, so derivative works were not at issue. A casual reader might thus conclude that these cases come to the same conclusion. But they are actually quite different. In Anthropic, Judge Alsup comes out swinging, arguing that model training is obviously transformative and writing that “to make anyone pay specifically for the use of a book each time they read it, each time they recall it from memory, each time they later draw upon it when writing new things in new ways would be unthinkable.” In Meta Platforms, in contrast, Judge Chhabria does say that “there’s no disputing that” large language models are transformative, he also says on page one of the opinion that “in most cases” such copying will be illegal, though he does not seem to have a deep theory for why this would be an exception. It appears that he relies almost entirely on the transformativeness factor even though a great deal of time is devoted to a market dilution theory of harm for the fourth factor. It feels similar to the famous Google Books case, Authors Guild v. Google 804 F.3d 202 (2nd Cir. 2015),4 where the use was so transformative, but also so useful, that it had to be allowed.

The other big distinction is on the effect on the market. In Anthropic, Judge Alsup focuses on the AI models’ inability to compete directly with their works. Instead, he almost derisively gestures towards the contentions of the authors that the “explosion” of competing works is a “generic” contention. He is surely right that it is speculative, but there is not a deeper analysis here, which leaves the reader wanting. In contrast, Judge Chhabria goes deep into this analysis, but the analysis itself falls short by failing to understand the zero marginal production cost of informational works. In one passage, he compares the effect of training an AI model on the life of President Lyndon B. Johnson to the market for LBJ biographies, saying that while Robert Carro’s Master of the Senate would be unaffected, lesser biographies probably would be. But what he forgets is that Master of the Senate has also decreased the market size of competing works too, including biographies that Carro relied upon! Judge Chhabria tries to create a new theory of market dilution, seemingly borrowing from trademark law, but this concept does not seem transferrable to me.

The main way that an AI model would compete with lesser-known, up-and-coming creatives would seem to me to be for works that do not yet exist. For example, the header article for this image was made by an OpenAI image generation model. It does not exist anywhere, but in the past perhaps I would have paid a smalltime artist to create it for me. There is a real market dislocation here with interesting social implications, but copyright does not protect works that do not yet exist, even though Judge Chhabria might want a new harm to be created for “indirect” substitution, contra Warhol’s clear directive that only direct substitution is protected.5

Both cases are quite different than the recent draft guidance from the United States Copyright Office. The prepublication Report was not exactly subtle in suggesting its authors’ view that training is not fair use, though it did steer clear of advocacy, saying that “it is for the courts to weigh the statutory factors.” The Report it did not come to any firm conclusions on any of the four factors, but it did make it clear that, in USCO’s view, any fair use analysis will prioritize factors one and four, and went out of its way to highlight public commentary that put forth arguments that AI training was not entirely transformative and that there would likely be a significant commercial effect on the works.

These cases are instructive when comparing to Thomson Reuters. In that case, Ross created headers very similar to the Westlaw headers for the purpose of creating a model that would form the basis of a competitive service after they approached Thomson Reuters about a license for the headers, which was rejected. This recalls cases like Fox News Network, LLC v. TVEyes, Inc., No. 15-3885 (2d Cir. 2018) and Associated Press v. Meltwater U.S. Holdings, Inc. 931 F. Supp. 2d 537 (S.D.N.Y. March 21, 2013), where if you are trying to use fair use to create a directly competing service—or, in the case of TVEyes, create a service that makes it just a little too easy to recreate a copyrighted work for the purpose of not paying for it. That is clearly not the goal here, where companies treat so-called “regurgitation” as a bug they work to eliminate. Indeed, in Meta Platforms, the authors could not get Meta’s Llama model to produce more than 50 words of their works.6

This is clearly a win for the model providers. These cases clarify that they can train on and digitize data as long as it is lawfully obtained. Although these are district court cases, because it hews so closely to existing precedent, I imagine that it is unlikely to be overturned and that we will see this become one of the seminal cases on the matter. This is exactly the result you would expect if you thought AI was a similar enough technology to other digital technologies to apply cases like Google Books, TVEyes, and Meltwater in an analogous fashion. The reason to believe otherwise would be if you believed AI was fundamentally different than digitization or search in a meaningful way. To me, I have been calling this the TVEyes divide in discussions with my colleagues precisely because this is where I think the dividing line will be: if the service is meant to create a generally useful tool or a clearly competing service. And given that almost all of the AI fair use cases have been dismissed on the substance, these cases might be all we get.

I believe that this is also a win for content owners, however. Just like with the early Internet, the various content industries like music publishing and movie studios began by fighting technological change. This worked poorly for them. Once they started to work with technology companies instead of opposing the very idea of technology, they began to make even more money than before. As an example, music streaming is more profitable than buying records ever was. While it is true that these cases put an implicit cap on the market for licensing content—specifically, at the cost of a single copy of the work—at the same time it makes the market reasonably sized and tractable, thus clarifying it and, I suspect, therefore catalyzing it.

What’s the upshot here? Copyright was always a somewhat limited option for artists to get paid by AI model providers given the strong fair use case. These cases put the spotlight on other IP rights, like trademark and likeness. They also encourage rightsholders to negotiation and practical solutions, like monetizing through AI-generated merchandise and derivative works. I have long been beating the drum, including on Nonobvious, that this was the path forward. I fear that this holding will encourage archiving at the expense of the open web, and that licensing terms for creative works may come with more limitations (unenforceable as they may be). But they will have also started to set the metes and bounds of a real commercial solution to this problem.

Just in time manufacturing was great for rightsholders. Perhaps just in time content will prove to be just as amazing.

Introduction

In recent years, China’s patent examination system has undergone a significant digital transformation, driven by the demand for faster and more efficient processes. One of the most notable changes is the integration of artificial intelligence (AI) tools across various stages of patent examination.

In this article, I’ll explore how AI is currently being used in China’s patent system, what the Chinese authorities have planned for the future, and what foreign attorneys—especially those representing overseas applicants—should know about these developments.

Current and Future Use of AI in China’s Patent System

As of 2024, AI is primarily used in the early stages of patent examination, including formalities review, prior art searches, and document classification. These tasks are structured and repetitive, making them a good fit for automation.

In the Sugon case, a leading high-performance computing and server manufacturer affiliated with the Chinese Academy of Sciences developed an AI platform for the China National Intellectual Property Administration (CNIPA). The system helps improve the efficiency of patent examination by automating the comparison and classification of technical documents.

However, at the substantive examination stage—where decisions about novelty and inventive step are made—AI still plays a supporting role. Human examiners remain the key decision-makers at this stage, although AI is expected to take on a larger role in the coming years.

Subscribe now

Looking ahead, CNIPA has made it clear that AI will become more deeply embedded in patent examination. This intent is reflected in strategic documents and public statements. For instance, CNIPA has issued trial guidelines for AI-related invention patent applications. Additionally, officials like Mr. Changyu Shen have stressed the importance of strengthening intellectual property protection in AI-driven sectors.

These developments suggest that AI will eventually be involved in more complex aspects of examination, such as analyzing novelty and inventive step. While the AI systems are still evolving, CNIPA is committed to making sure this transition happens smoothly and with careful oversight.

Practical Implications for Attorneys

For patent attorneys in China, the rise of AI presents both benefits and challenges. On one hand, AI helps increase efficiency by automating routine tasks like prior art searches and document classification. On the other hand, it introduces new challenges, especially when Office Actions are based on AI-generated prior art comparisons.

A common trend now is the increasing number of Office Actions that cite numerous prior art documents. These citations are often selected by AI algorithms that may not fully capture the context or relevance of each reference. This can make it harder for attorneys to craft accurate responses. Additionally, Office Actions are now being issued faster and include more data, which requires skilled professionals to manage effectively.

A good practice is to cross-check each AI-cited reference manually to ensure it is technically relevant. Maintaining a balance between AI-assisted insights and human judgment is key to effective prosecution.

For foreign attorneys, understanding this shift is important. While AI-generated Office Actions are currently more common in domestic applications, it’s likely that foreign filings will see similar changes in the near future. Attorneys representing overseas applicants need to be prepared for this shift and adapt to the evolving system.

A Changing Role for Human Expertise

One of the most surprising implications of AI in patent examination is how it is changing the role of human expertise. AI is not merely accelerating the process—it is reshaping the way patent attorneys engage with their craft.

Patent attorneys must now go beyond traditional legal reasoning. They need to understand how AI systems select and interpret prior art, and they may need to challenge AI-generated conclusions. In a sense, they will need to learn how to “argue with the AI.”

In a recent case I handled, the AI system flagged a Chinese utility model as prior art for a software-related invention. The cited document included some overlapping keywords and technical terms, but on closer review, the actual solution in the prior art was completely different and lacked the key inventive concept in our claims. Initially, this confused the client, but we were able to clearly demonstrate to the examiner—through a side-by-side analysis—that the AI’s citation was not relevant. The human examiner agreed and removed the reference in the next Office Action.

This kind of situation shows that while AI can help surface documents more quickly, it doesn’t always grasp the true technical essence or context. As a result, attorneys who can combine technical knowledge with strategic thinking are becoming even more essential. In particular, they must be able to distinguish between valid AI-generated objections and those based on irrelevant or misleading data.

Conclusion

The integration of AI into China’s patent examination process is more than just a temporary experiment—it’s a fundamental shift in how patents are examined. While the efficiency gains are clear, both domestic and foreign attorneys need to be ready for an era where human expertise and machine learning intersect. The future of patent examination will require not only legal skills but also an understanding of how AI works and how to navigate its challenges. Those who can navigate both legal frameworks and intelligent technologies will define the next generation of patent professionals.

Weekly Novelties

• DeepSeek, the Chinese AI wunderkind, turns out to have filed for at least 19 patents (IAM Media)

• The Federal Circuit overturned an Apple PTAB win against Qualcomm, finding that the written materials evidentiary restriction was very strict and did not allow the PTAB to loop in the applicant’s own disclosures (IP Watchdog)

• The Federal Circuit held that merely applying machine learning to a new domain is not subject matter eligible; surprisingly, this was not an obviousness or enablement case (Holland & Knight)

• A jury awarded over $100 million in damages against Elon Musk’s X Corporation, formerly known as Twitter; this will likely make him even less of a fan of patents (Kera News)

• Nokia’s Q1 report showed significant changes in its licensing revenues (IAM Media)

• USPTO accelerated the issuance timeline for patents, aiming to reduce it from 3 weeks to 2 weeks (National Law Review)

• EPO President Antonio Campinos said that US instability induced by tariffs introduces an innovation opportunity for Europe (Euronews)

• An analysis finds that some numbers are more popular for trademarks than others, like 76 (Fast Company)

• China’s United Trademark Group, one of the world’s largest trademark licensors, invested in Xcel Brands to expand its global reach (Women’s Wear Daily)

The idea of a registration system gets to a surprisingly longstanding debate as to the right balance between ex-post and ex-ante patent evaluation. Some scholars, like Mark Lemley, have long argued for “rational ignorance” in the patent office and a “soft review” of patents, though his famous article stops short of arguing for a true registration system. Some have argued that, with the complex mix of ex-post review, that the US is already nearly a de facto registration system anyways.

What is missing from this debate is evidence. What do we know about examination today and its costs and benefits? And are there other ways the patent office could be more efficient?

Registration pending?

What is a registration system? In a patent registration system, patent applications are reviewed primarily for meeting the procedural requirements for registration—typically formalities—with other issues like validity handled ex-post in the court system. In other words, imagine a patent system where getting a patent hardly required an attorney and where you didn’t know whether the patents even protected your IP rights until they were litigated.

Generally, the arguments for registration systems tend to focus on arguments of efficiency. Overall, the main argument for a registration system is that the current examination system takes too long process patents and is too expensive, especially since one review by ex-judges found that valid claims were more likely to be rejected than invalid claims allowed. There are also arguments that registration systems promote equity, particularly because registration systems are cheaper for pro se applicants to access, both because attorneys can be expensive and because attorney quality has a measurable impact on the likelihood of affecting the likelihood of a patent getting granted (to say nothing of the scope of the claims). There are also some more exotic arguments about efficiency. One article from 2003 argued that a registration system would theoretically incentivize patentees to better pursue claims by reducing administrative costs and shifting the burden to opinion of counsel, similar to self-certification in building permits.

Patent registration systems are quite rare, which makes them hard to study—when I first heard of registration systems without examination I felt like a fish who had just heard about water and some strange place without it. The few countries with registration systems tend to be ones that have low volume, like Nigeria, which saw fewer patent applications in 2020 than 2012, or South Africa, which is actually now introducing an examination system. One other surprising example is the United States itself, which started its patent system as a registration system before abandoning it in 1836. The United States originally1 adopted a British-style patent registration system for several reasons—patents were viewed as a property right and there was an express goal of making patents cheaply available to even those with humble means—but the “disastrous experience,” as Berkeley’s Robert Merges described it, of excessive litigation of the resulting low-quality patents caused Congress to eventually abandon registration. When Senator John Ruggles created the patent office in 1836, he also effectively invented the patent examiner, and for a while it was a popular job—some were even celebrities!2

Subscribe now

Given the paucity of registration systems, there aren’t any studies comparing examination and registration systems. That doesn’t mean that we are flying completely blind with respect to evidence, however. The data doesn’t look good for registration systems. The only serious quantitative treatment I could find was a study from Duke Law School by Michael Frakes and Melissa Wasserman in 2019 found that the cost of ex-post litigation was higher than the cost of increased examination and that, according to their model, more examination would more than pay back a lack of litigation costs. If we think of registration as a system with zero examination costs, then this study implies that we would be going in the wrong direction with registration. The Lemley argument for rational ignorance relied on the simple observation that most patents do not appear to be used, but the Frakes and Wasserman study accounts for the difference in costs to come to a more complete view.

Whether or not adopting a registration system is a good idea, it is probably not something that the Administration could pursue on its own without Congressional action. 35 U.S.C. § 131 says that “The Director shall cause an examination to be made of the application and the alleged new invention.” The law, in other words, is clear that some level of examination is required. While USPTO can waive its own rules, change the MPEP, and loosen its examination standards, it must still comply with the statutory requirements, which seem to require some level of examination for the non-procedural elements of a patent.

In other words, a registration system probably wouldn’t help make the patent office more efficient and probably can’t get done.

So what can? In the past, companies like Cloudflare have started new initiatives like prior art bounty programs. Some have proposed the use of new technologies like blockchains to make patent data more open and therefore easier to register relevant prior art. In general, those who want to reform the patent system should not dispair. New reforms can take hold—ex-parte review was only introduced in 1980 and the post-grant review aspects of the PTAB were introduced in 2012 by the AIA.

When it comes to technology, one idea that has been floated around is the use of AI-assisted substantive prior art search and examination. While some have mocked this idea, it is actually the single most compelling thing examiners can do to enhance their efficiency. Although the previous administration banned the use of generative AI by patent and trademark examiners, we may see a rescition of that memo (at least for specialized, USPTO-approved tools).

Examiners are overworked, but an analysis of the how they spend their time shows that a majority of their efforts are spent on generating the first office action. According to a USPTO study, nearly 50% of all examiner time on a case are spent on just conducting prior art searches and drafting the initial rejection letter, about 25% each. These are both tasks that AI excels at. It is very feasible for AI to conduct an initial, comprehensive search based on the claims to find, rank prior art, and provide an initial mapping against the claim elements much faster than a human can. While examiners will need make the final determination on their own and may conduct additional searches, this can get the easy stuff out of the way and allow the overal caseload to get cleared faster. AI could also provide a first cut at other types of analysis, like identifying lack of antecedent basis and recognizing new matter in claim amendments. AI also excels at drafting, accessing boilerplate, and querying a corpus of guidance, like the MPEP and the law. This means that once they have their arguments, examiners can spend less time drafting and have AI turn their arguments into clear, well-formated, standardized rejection (or allowance!) letters, with AI pulling in the appropriate citations to prior art or the MPEP. As AI gets more advanced, it can even help junior examiners take advantage of strategies more commonly used by senior patent examiners, like examiner amendments, by suggesting when they may be appropriate.

The need for a more efficient search and examination process clear, but a registration system will hurt more than it helps. But for those at the Trump Administration who read Nonobvious, it is time to take the idea of AI-assisted examination seriously, especially at the front end of the process.

Bring AI to your patents with Edge

Prior Art

Several weeks ago, we covered USPTO’s overarching AI strategy, observing that it wasn’t much of a strategy at all. Since then, others have echoed this opinion, including Dennis Crouch at PatentlyO, who referenced Nonobvious in a post. Last week, Acting Director Coke Morgan officially rescinded the short-lived AI strategy. With a new incoming director and an Administration that has many staff who care deeply about patents, there is a strong opportunity to bring a real AI strategy to USPTO. As I argued in this week’s main piece, search and substantive examination would be a great start.

Weekly Novelties

• Two different George Church startups have run into issues over IP conflicts. This is a fascinating story of conflicts of interest, unclear IP overlap, and what happens when you have one particularly prolific inventor at the center of too many startups (Fierce Biotech)

• In no surprise, an appeals court held in Thaler v. Perlmutter that DABUS could not autonomously receive copyright protection. More important for practitioners, Thaler’s later-asserted human input was waived because it was not asserted in the original filing with the Copyright Office (Professor Edward Lee’s writeup)

• White and Case joined the defense team for ROSS in its AI copyright lawsuit against Thompson Reuters, and asked the judge to certify copyrightability and fair use for an interlocutory appeal (Law.com)

• Secretary Lutnick fired all members of the USPTO Patent and Trademark Advisory Committees (PPAC and TPAC). Although both committees serve at the pleasure of the Secretary, and the committees are supposed to regularly rotate, it is unusual to clear out everyone at once (PatentlyO)

• Oral arguments were heard this week in Google’s appeal against EcoFactor, arguing that expert testimony was improperly permitted. The oral arguments were not favorable for EcoFactor, though it was not immediately clear how the court would rule or how narrow the ruling would be (Larkin Hoffman)

• As the pharma patent cliff continues, pharma braces for the consequences. Novartis announced layoffs of 427 employees as its patent for Entresto (North Jersey)

• Why has the Biosimilars Act of 2010 not had the same impact on biosimilars that Hatch Waxman had on generics? One intriguing argument—biomanufacturing patents that may cover the manufacturing of many biosimilars (Nature Biotechnology)

Try Edge now!

To start, Edge tracks three types of errors:

• Antecedent basis errors. Edge not only tracks obvious errors and obvious passes in the claims but also ambiguous cases

• Patent profanity throughout the application

• Reference errors, like referring to a figure that doesn’t exist

In our internal testing, the Edge proofreader meets or exceeds the performance of other proofreaders. On top of that, Edge’s user interface is easy to use and understand.

To access the Edge proofreader, click the “Proofreader” button on the right, next to “references.” No downloads and no fiddling with various clouds. Just a proofreader that works directly in Edge.

From there, you can view every identified error, or click on a category to filter only errors in that category.

If you have multiple errors of the same type, like multiple antecedent basis errors for the same phrase, you can see that they are grouped together with a count. You can click on the error and use the arrows to filter between them.

Get a free proofreader

Our proofreader is going to get smarter over time, adding in more error types, improving the sophistication of error checking, smarter error sorting, and correction suggestions. You can’t find a proofreader like this anywhere else and it’s just going to keep getting better.

The US Patent and Trademark Office (USPTO) has dedicated countless hours to an increasingly urgent task: determining how much human involvement is necessary to qualify as an "inventor" under patent law. The agency’s focus is understandable, but the rapid pace of technological advancement means that soon, this will no longer be a matter of nuance or interpretation—it will be a necessity. 

Imagine a not-too-distant future where artificial intelligence (AI) systems develop a new compound that could cure cancer, and no human can be fairly identified as the inventor. What happens next?

Subscribe now

Under current patent law, the answer is simple but troubling: if no human can be listed as the inventor, no one owns the patent. This presents a significant problem. In the absence of clear ownership, no company or individual will spend the enormous amounts of money necessary to test, refine, and bring this groundbreaking treatment to market. The result? Potentially life-saving innovations could languish in the dustbin of history, unused and forgotten.

The root of the present ownership issue lies in an outdated legal framework. The Patent Act, as it stands, requires an "inventor" to be human—which makes sense if we assume that only humans can invent. But today, AI systems are already contributing to cutting-edge advancements in fields as diverse as drug discovery, materials science, and cybersecurity. And tomorrow, those same AI systems could make discoveries entirely on their own. When they do, our laws will be unprepared to handle the implications.

This is not a hypothetical scenario; it is an immediate reality. If we don't act now, we risk stifling innovation precisely at the moment when AI has the potential to transform entire industries and solve some of humanity's most pressing problems. 

Try out Edge's AI patent system

The Constitution grants Congress the power to promote the "progress of science and useful arts" by securing for limited times the exclusive rights of inventors. For obvious reasons - it was 1787 after all - nowhere does it say that those inventors must be human. In fact, as Justice Sandra Day O’Connor famously noted in the Feist v. Rural Telephone Service Co. 499 US 340, 349 (1991) case (often referred to as the "White Pages" case), the “primary objective” of Article I, Section 8 is not to reward creators for their own sake but "[t]o promote the Progress of Science and useful Arts." Indeed, the patent system exists in part to encourage innovation and ensure that society benefits from it.  Motivation is just one goal - and a very good one.  But the other is much bigger and recognizes the clause’s primary objective that is broader and benefits us all: “To promote the Progress of Science and useful Arts.” At the end of the day we have to ask: what does society get?  The answer to this question is our lodestar.

There is nothing in the foundational text of our patent law that prevents us from recognizing non-human creators. The Constitution itself is flexible enough to allow Congress to adapt the law to meet the needs of a changing world, just as O'Connor emphasized in the context of copyright law.

Critics will argue that allowing non-human inventors will create a host of legal and ethical challenges. Who would own the patent—the AI itself? Its developers? The company that trained the algorithm? These are valid questions, but they are questions we can address through thoughtful legislation. 

Thankfully, there are signs that at least some in Congress are paying attention to issues with our patent system. Already, there are two recent Senate initiatives tackling other perceived shortcomings in the patent system, the Promoting and Respecting Economically Vital American Innovation Leadership (PREVAIL) Act and the Patent Eligibility Restoration Act (PERA). These bills have both shown serious potential to pass Congress, which signals momentum for meaningful reform in crucial areas.  Congress must add AI inventorship to that list to ensure that our laws are forward-thinking enough to embrace the future, rather than clinging to the past.  

The clear Congressional impetus to pass legislation, like the PREVAIL and PERA, should trigger broader discussions that go beyond a single political moment. While no one has all the answers, comprehensive hearings would allow a diverse range of stakeholders—small inventors, large corporations, academic researchers, and legal experts—to highlight their respective views. Thanks to the DABUS series of cases, this crucial issue is being handled by the courts based on what the law is today. What we need to do instead is discuss what should be.

As to my view: we must amend the Patent Act to reflect the reality of AI-driven innovation.  Congress can get this done in a thoughtful, non-partisan way. Otherwise, we risk standing idly by as life-saving inventions go undeveloped simply because our legal system couldn’t keep pace with the technology that created them. The ultimate goal of the patent system is to serve the public good. Let’s make sure it does.

Jeffrey E. Ostrow is the former Chair of the Intellectual Property Department at Simpson Thacher & Bartlett LLP.  He currently consults on strategic IP issues companies face and teaches at Berkeley Law School, where he developed a class called the “Business of IP” which examines how intellectual property issues increasingly shape and drive real world decision making.

Prior Art

Last year, I covered and explained the fight between Scarlett Johansson and OpenAI over a voice product that she alleged sounded eerily similar to her character in the movie Her. (Although she claimed to have retained lawyers, no case was ever filed.) In the article, I observed that copyright seems like a less-clear example for entertainment applications than likeness and trademark. Johansson now seems to be the poster child for AI issues for Hollywood. This week, a viral AI-generated video went public of “celebrities” wearing obscene t-shirts in protest of Kanye West’s Nazi t-shirt website. Though she approved of the message, Johansson called for federal AI likeness legislation. This is a fight that is not going away. Now, I don’t think that anyone is going to think that this is real, or that the founders of Google and Facebook sat down for this video. But you could imagine a much more believable video, especially now that AI video generation is getting so good.

Weekly Novelties

• Vaishali Udupa, the commissioner of patents for USPTO, resigned (IAM-Media)

• The Federal Circuit upheld the PTAB’s IPR authority over expired patents. See Apple Inc. v. Gesture Tech. Partners, LLC, 2025 WL 299939, *2 (Fed. Cir. Jan. 27, 2025) (link)

• USPTO issued clarifications for its new IDS fee schedule. In short, submissions in reexamination proceedings are exempt from IDS size fees while filing benefit claims in a reissue will result in new continuing application fees (JD Supra)

• In gambling news, a lawsuit between two roulette companies, Evolution and Light & Wonder, came to an end as the roulette patents were held to pertain to abstract ideas. Game patents are not illegal per se, but typically relate to technical aspects; game mechanics are very gray post-Alice (Next.io)

• The European Commission withdrew its controversial SEP legislation, saying that there was “no foreseable agreement” to be reached. However, it left the door open to pursuing different legislation in the future (JUVE Patent)

• Relatedly, the European Commission also dropped legislation to regulate AI that would have allowed consumers to sue over fault or omission of an AI provider (Reuters)

• Trina Solar sued Canadian Solar over patents relating to tunnel oxide passivated contact cells, pursuing $147m in damages (PV Magazine)

Hope is not a Strategy (and No Details is Just Hope)

What is most important is what is missing: a clear, fleshed-out technology strategy. USPTO has had a lot of egg on its face in recent years when it comes to technology, from data leakage to DOCX issues to the debacle that is the Patent Center. So USPTO’s Focus Area 2, which is to “build best-in-class AI capabilities,” unsurprisingly raises some eyebrows. Is USPTO capable of building such priorities (and why isn’t it Focus Area 1)? The first set of sub-priorities all relate to computational infrastructure, but there are no specifics about what type of infrastructure is needed. If they are buying Nvidia GPUs, they will need to increase fees to pay for it!

But perhaps most concerningly, despite having had over a year to create this strategy, there is no “what” and no “how” in even a basic sense. USPTO promises to “identify” areas of opportunity, but shouldn’t it have found some of those areas already? Some obvious ones would include AI-assisted prior art search for examiners, AI-assisted form filling, and AI-assisted prosecution to bring down its worsening pendency timelines. Perhaps USPTO could have given some clarity as to its preferred approach—will it try to fine-tune its own LLM, or perhaps create a startup program to bring in the best and brightest? All of this is an exercise left to the reader, and it leaves that reader with an unfortunate lack of confidence that USPTO has the capability to pursue this part of the strategy. This particular area needs more fleshing out, particularly since Focus Area 4 emphasizes building up AI training and expertise within USPTO without clarifying what type of training or expertise—does USPTO want to train examiners well, or does it want to hire its own AI developers?

As a general matter, the strategy is very light on details. USPTO’s Focus Area 1 focuses on studying and publishing on the interplay between AI and IP policy, including issues like changes in filing volume, subject matter eligibility, and more. USPTO has already been publishing on this topic, including AI invention examples and guidance on inventorship; it also has a robust econometrics-driven research division. USPTO already engages with policymakers to promulgate its preferred policies and already has several outreach programs to help increase inventorship. Similarly, Focus Area 5 says that USPTO will engage with stakeholders, but it already consults with other agencies and global IP offices on matters like AI. Therefore, it is not clear what is changing as a result of this document.

One of the only details in this strategy is that USPTO, under Focus Area 3, is its interest in promoting “responsible” AI use in the community. Here USPTO emphasizes privacy, upholding civil liberties, and avoiding discrimination. While these are all important values, it is not clear how they are impacted by AI in the IP community specifically; much of this language seems to be simply lifted from President Joe Biden’s executive orders on racial equity. Unfortunately, USPTO specifically endorses an AI standard called NIST AI RMF. This is unfortunate, partially because it is too early to endorse any particular AI standard in this fast-moving space and partially because the NIST standard is vague and not gaining much traction.

Try AI patent drafting

But is this AI strategy already dead? Several of the strategy items are built on Executive Order 14110, which is better known as President Biden’s AI executive order, as well as Executive Order 14091. In addition to directing federal agencies to appoint a “Chief AI Officer” and expressing an interest in promoting competition, this executive order emphasized the importance of preventing discrimination through AI systems certain reporting requirements. Those executive orders were rescinded by President Donald Trump within hours of his assuming office. Presumably, this strategy will need to be changed as a result. USPTO should take this opportunity to put out a more ambitious, and specific, plan focused on technical and training milestones with key performance targets.

As a final note, USPTO’s document bills itself as a strategy document more than a policy or position document. This is largely accurate, though it will need to pick up the pace on policy to match its goals under Focus Areas 1 and 5. The US Copyright Office has been studying the topic of AI and has been putting out position papers. The most recent paper, on AI and copyrightability, was put out on January 29, just two weeks after USPTO’s strategy document came out. And other countries aren’t standing still. Japan, already a leader in AI policy, is exploring the use of AI in streamlining its prosecution process.

Weekly Novelties

• DABUS, which has pursued the cause of AI inventorship, has lost another case, this time in Japan, where its Intellectual Property High Court (similar to the Federal Circuit) upheld a Tokyo district court decision holding that AI cannot be an inventor on a patent (Mainichi)

• USPTO continues to be in disarray over the new Return to Office orders; so far, PTAB judges are required to come back to work, and supervisors may have to as well, though so far examiners are protected under their collective bargaining agreement (MLex)

• A recent analysis found that the Netherlands has grown rapidly in oncology patents and is now the #3 patent filer in the space in the European Union (NL Times)

• Last week in Steuben Foods v. Shibuya Hoppman Corporation, the Federal Circuit inched closer to abolishing the reverse doctrine of equivalents; although they found a compelling argument that the 1952 Patent Act eliminated the concept, they did not need to address the issue in this case, so RDOE lives to fight (but not be used, ever) another day (JD Supra)

• Two patents by 10x Genomics were invalidated at the PTAB in the latest salvo of a legal battle against Parse Biosciences (Bloomberg)

Today we are taking a break from all of that to announce a feature that has been one of our most-requested features.

The figure editor now supports squiggles.

Edge now supports a variety of lead lines, including squiggles and underlined numbers. We will also be rolling out support for other common lead lines, like brackets.

Try Edge today

We also support squiggles in flow charts.

Although including the number in the figure is more common, we know that many of our customers prefer a squiggle there as well. For those who prefer an external lead line in flow charts we now support that as well.

Schedule a demo

Keep posted; Nonobvious will be coming back with more content on patents, IP, AI and the law. And we will also provide exciting updates on Edge.

The ABA has now issued a formal opinion on the matter, ABA Formal Opinion 512. This is coming just in time as Bar Associations all over the country are starting to issue (generally very favorable) guidance on the use of AI tools in the legal practice. The guidance from the ABA was generally favorable but somewhat vague; in many ways it was the same with USPTO. This week we will dive in to the guidance and what it means for practitioners, and where there is still a need to tighten it up.

Overview and Comparison to USPTO

Competence

Lawyers have a duty of competence to their clients. This doesn’t mean that lawyers have to win every case but that they must do their best and not make obvious mistakes. In the real world, lawyers engage in some level of delegation all throughout their jobs. They use associates and paralegals, send overflow work out to other attorneys, and even use software like spell-checkers, search tools, and statistics. A duty of competence is therefore, implicitly, a duty of oversight, both in the use of tools and in understanding “the benefits and risks associated with the technologies used to deliver legal services.” For the ABA, this means the following:

• Understanding how AI tools work as a user; this does not require becoming an AI expert. The focus is on knowing the strengths and weaknesses of the tools

• Keeping up to date on the fast-moving field of AI (one easy way to do that; subscribe to Nonobvious!)

• Reviewing output for accuracy, bias, quality, and more; there cannot be “uncritical reliance” and lawyers may not “abdicate their responsibilities” specifically when “professional judgment” is involved

When it comes to competence, the ABA is shockingly forward-thinking and understands that AI may become required to provide competent council in the future. The guidance observes, for example, that one could not competently serve clients without sending email, creating electronic documents, or conducting online searches. What is not clear here is what level of materiality is important for competence; that term does not appear anywhere in the guidance.

The guidance also leaves unclear what the baseline is to avoid “abdicat[ing] their responsibilities.” I am generally critical of guidelines that treat machines more harshly than humans. If the machine makes mistakes as often and of a similar type as an associate, and the mistakes are of the same level of materiality, is a lawyer satisfying their ethical obligations by exercising the same level of care that they ordinarily would with a junior employee? Or does a lawyer have a higher standard—implicitly, perfection—when they use AI assistance?

USPTO makes the same observation, highlighting the “duties of competence and diligence.” Unlike with the ABA, USPTO does specify that the practitioner must act with “reasonable diligence,” though it still doesn’t say what that means. USPTO does require practitioners to “ensure that all statements in the paper are true” when it comes to facts, “confirming the accuracy of all citations to case law” and other references, and “ensure that all arguments and legal contentions are warranted by existing law.” To some extent, this does fly in the fact of how practitioners do their jobs today: patent attorneys do rely on boilerplate written by other people in their firm without checking every example in a definition, for example; they may also rely on information in issued patents without checking everything about it. When does this cross the line? That is as unclear.

Subscribe now

On this issue, the ABA has a funny, short section on “meritorious claims and contentions and candor toward the tribunal.” In essence, there have been so many stories now of litigators in particular using ChatGPT-powered tools that have created false citations or spurious arguments. It is a particular reminder that these obligations are heightened when making statements before tribunals to cover even accidental misstatements.

Confidentiality

Confidentiality is one place where every lawyer asks questions, and rightly so. The ABA and USPTO have similar things to say on the matter. Essentially, lawyers must take “reasonable efforts to prevent the inadvertent or unauthorized disclosure” of client information. For IP, where disclosure can destroy patentability, this is particularly important. USPTO and the ABA are drawing from the same set of model rules here, so the guidance is relatively similar. The ABA does not set an unreasonable diligence standard here; lawyers should read the terms of service and privacy policy, and may want to talk to an outside expert in cybersecurity or IT. But the focus is on understanding the policies, not on requiring a full forensic analysis. This is quite reasonable and a good best practice.

The ABA takes a position that is balanced overall, allowing lawyers to assess the likelihood of disclosure, the sensitivity of the information, the difficulty of implementing safeguards, and even the potential impact of those safeguards on the quality of client representation. Many of the requirements—such as access control both in and outside the firm—are not unique to AI tools. “Self-learning” AI tools get a special shout-out for concern given the potential for confidential information to make become embedded in a general model that may be accessible to other customers of an AI service.

Although we will talk more about disclosure and consent in the next subsection, the ABA is quite clear that self-learning systems require client consent. It is important to note that this line does not apply generally to AI tools; just ones that are self-learning. Furthermore, the way that the ABA uses “self-learning” suggests that the particularly disclosures are required for training for generative systems only, not other categories like data aggregation for statistical purposes or discriminative models used for classification. Surprisingly, though, the ABA raises the bar by saying that “merely adding general, boiler-plate” to engagement letters is not sufficient, but it does not say what is. For lawyers who are making this decision now: self-learning is not the default for machine learning. Generally, consumer services like ChatGPT are self-learning; enterprise systems like Edge generally are not. You should ask, but know that self-learning is an active decision made by a software developer that requires work and maintenance. And you can always review cybersecurity certifications to help bolster your confidence, like SOC 2 Type II.

Communication

ABA and USPTO are most unclear on the topic of when to disclose the use of AI to clients. Both rely on the rule that lawyers must “reasonably consult with the client about the means by which the client's objectives are to be accomplished” but neither specifically require a duty to disclose the use of such tools. While the ABA says that disclosure is not always required, it may be, and it depends on “the facts of each case.”

Subscribe now

There are a few clear bright lines. If a client asks whether a lawyer is using AI, they must answer honestly. And if the use of AI is germane to a justification for a fee that must be provided as well. One thing that is not quite clear is whether disclosure is required by the tool. The ABA suggests that disclosure is required when the lawyer “proposes to input information relating to the representation” into an AI tool. But what does that mean, and why is that required if the lawyer has already done their diligence on the tool per their obligations relating to competence and confidentiality? Lawyers provide information “relating to the representation” all the time into non-AI tools. Confidential workflow information is entered into Clio, with documents saved on Box, and researched on Lexis Nexis. You do not have an obligation to disclose those. And isn’t some level of disclosure required to use the tool—meaning that this would be a backdoor requirement to all uses of AI? The most interesting example of a required disclosure is when AI is used to help render a judgment rather than merely producing output. But would, say, a statistical ranking system require disclosure if the lawyer is making the decision? What if the ranking is only used to sort the items of interest for efficiency purposes? And which technologies would be required to count as “AI”?

What is a practitioner to do with this advice? While USPTO does not require you to disclose the use of AI tools to USPTO, it does require you to stand by your signature, and to disclose when a claim was not touched by a human inventor; it also puts practitioners on notice that requests may be made about the use of AI, specifically around inventorship. And while the USPTO guidance is vague and thus relatively permissive, the ABA rules leave very little that would not be required to disclose, possibly, but without defining well which types of tool use would require disclosure.

This is the area that will need the most clarification, and likely, revision.

Firm management

The most interesting subsections of the guidance—which shows how deeply the ABA thought about this problem—is areas that fall under the rubric of firm management (though the ABA does not group them these way). They include fees and supervisory responsibilities.

On the managerial side, ABA advises that law firms need to establish training programs, clear AI policies, and supervisory rules that take into consideration the AI tools they are using. The ABA highlights that lawyers have particular duties to nonlawyers in particular. Much of what is there is standard practice when using vendors and subcontractors, like reference checks and understanding their policies. If you have followed our series on how to buy software, you will largely be following these practices already.1

Subscribe now

The other is fees. First, lawyers must clearly communicate any AI related surcharges clearly and ahead of time. This is simple and just good client management. Note that although the ABA says that this practice is not forbidden, though the ABA also says that “overhead” may not be charged back to clients; this is an ambiguity that will need to be resolved later (though the market will likely handle this; clients may not tolerate this for long). You also must be honest: if AI makes you faster, you must still be honest in your billing. We will see if there is increased ethical enforcement on this matter. The move to flat-fee billing for more matters will likely make this irrelevant for many fields.2 However, it is interesting that the ABA is taking a stance here (at least, ethically) at the efficiency conflict of interest inherent in the billable hours model. The ABA’s answer to lawyers is: deal with it. This is particularly notable given that the ABA has acknowledged that AI may be required to provide competent counsel in the future; arguably, this would create an ethical obligation for efficiency.

Weekly Novelties

• Y Combinator held a policy event on AI, featuring FTC Chair Lina Khan and California State Senator Scott Wiener; the talks by the speakers are online (watch on YouTube, report on LinkedIn)

• In the newest of a string of bills attempting to overturn the past two decades of SCOTUS patent jurisprudence, two senators introduced the RESTORE Act, which would undo eBay v. MerExchange, which was criticized for making it too difficult to get an injunction (IP Watchdog)

• Amazon sued Nokia for patent infringement over networking and virtualization patents, a rarity for a company that normally sees itself as the plaintiff (The Register)

• Sanofi sued Sarepta over manufacturing patents in the gene therapy space; cheap manufacturing is the key to reducing the cost of gene therapies, which are often 7-figures in cost (Fierce Bio)

• Qualcomm signed a large licensing deal with Honor, an OEM (IAM)

• Western Digital was hit with a $262 million jury verdict in a hard drive patent case (Reuters)

• The Federal Circuit held that the PTAB did not overreach in a voice technology patent case, though it allowed the patent holder to preserve its claim construction arguments (Bloomberg Law)

Although the Republican Platform does not address patents, the Heritage Foundation’s Project 2025, which is the unofficial but influential Republican shadow platform prepared by former Trump aides, does have a lot to say in its section on the Commerce Department. This is particularly interesting given that there have been several important patent bills introduced this term, like PERA and PREVAIL. So perhaps it is unsurprising that the detailed version of the plan has a lot to say about how it would approach these issues, as well as other priorities. This week in Nonobvious, we’re going to go over what the Republicans may have planned for the patent system.

Subscribe now

What Heritage Has Been Cooking

What Project 2025 makes clear is that conservatives have decided to make patents a priority.

The first main proposal is to review patent eligibility requirements and PTAB rules. While Project 2025 doesn’t endorse PREVAIL or PERA, this is nonetheless evidence that Thom Tillis (and perhaps former Trump Administration officials) have the ear of leading conservatives. While these proposals are bullet points without extreme detail, there are still some important indicators as to where the authors are thinking.

When it comes to patent eligibility, the authors (who explicitly mention Section 101) indicate that their focus is on enhancing American leadership in key technology areas. Three areas they highlight are quantum computing, 5G, and AI. Regular readers of Nonobvious will know that the 5G patent wars are very strong; nearly every week, our Weekly Novelties section includes a major 5G patent lawsuit, SEP dispute, or cross-licensing agreement. Similarly, quantum computing does not seem to be hampered by 101. AI will be much more affected by PERA, but it is worth noting that AI patents are being issued at a blistering rate. I was disappointed to not see biotech get a mention, particularly given that Mayo was the opening shot in the war on eligibility and that it is one of America’s leading areas of technology relative to China.

When it comes to PTAB, Project 2025 emphasizes the “speed and transparency” of IPRs. This indicates that they don’t fully understand the purpose of PREVAIL. Although more speed and transparency are always good, the main complaints with the PTAB have to do with the ease of raising issues, even from uninterested parties at any point in the life of the patent, which standards are applied during IPR (as opposed to litigation), and whether the judges make appropriate decisions. There is also a question of whether the PTAB has made it easier to fix otherwise invalid claims. None of these have to do with speed or transparency; the PTAB is much faster than litigation already and the public record for IPR is much better than federal courts. This suggests to me someone at the Heritage Foundation spoke to someone in Congress and wanted to get a Project 2025 endorsement for PTAB reform but that the staff did not fully care to understand what needed to be fixed.

The second interesting aspect of Project 2025 is that it emphasizes the international trade aspects of patents. In addition to promising to enhance trademark counterfeiters and infringers (although there is no specific proposal), Project 2025 takes a strong stance on supporting “like-minded countries” to rise to WIPO leadership positions. This is good; usually, American administrations focus more on the World Bank. Project 2025 also talks about strengthening international IP rights. Previously, this was a big part of American foreign policy. One of the main components of TPP, the trade treaty opposed by President Trump, was to enhance patent rights. While Republicans cannot get back into TPP without losing face, perhaps they can start by creating a subsidiary convention that only adopts the IP provisions as a way to easily make progress on this promise.

Interestingly, Project 2025 takes a stand against patent waivers for COVID-19 vaccine patents and in general, calling out the WTO’s Trade-Related Aspects of Intellectual Property mechanism in particular. This is the only bone thrown to biotech in the entire proposal, but it is a big deal. The Biden Administration has taken opposing Pharma patent strategies as a major tentpole position (more on that below). Although Project 2025 does not go so far as to support Pharma patents, this is the best indication that the Project 2025 team supports strengthening intellectual property rights, particularly because it is relatively niche but high-impact. This will be particularly important as the 2028 patent cliff looms. Already, leading Democrats like Bernie Sanders are taking aim at semaglitude patents as “junk” patents, which shows that the war against Pharma patents will not stay limited to the emergency justification levied for the pandemic.

The last significant proposal is a hairbrained scheme suggesting folding USPTO into another department. The top choice of the Project 2025 authors is to merge USPTO into the OMB as a Performance Based Organization (“PBO”); in the alternative, the authors propose merging USPTO with NIST and closing or spinning off all “non-mission-critical research functions.” This first proposal is wrongheaded and nonsensical. Also, strangely, they do not suggest merging with the copyright office, which would merge all IP under one roof; in fact, Project 2025 has nothing to say about copyright at all.

First, a brief history on PBOs. In the 1990s, a key Clinton Administration priority was increased government efficiency as part of the triangulation strategy. One initiative that got a lot of attention was the idea of a PBO. A PBO is an agency designation that sets performance targets for an office, particularly around its finances. Ultimately, although several dozen agencies were originally proposed, only three were ultimately selected in the final legislation, one of which was…USPTO. So it is odd that Project 2025 proposed turning USPTO into a PBO when it already is one. Furthermore, USPTO does not engage in much research, so there is not much to spin off. Merging with NIST would be odd, but at least it is a technical agency that is connected to USPTO through CFIUS; merging into OMB, which is a budgeting agency, makes no sense at all. This seems to stem from the desire of conservatives since the 1990s to eliminate the Department of Commerce. Ultimately, USPTO is stretched as it is; putting it under the purview of non-experts would hamper it further.

A Coda on the Democratic Platform

As I got around to writing this week’s edition, I was hoping that I would be able to write next week’s issue on the Democratic Party’s platform for patents. (Hey, it’s always easier when the news cycle tells you what to write about!) Unfortunately, the Democratic Platform has almost nothing to say about patents and intellectual property. So I will briefly summarize them here.

Subscribe now

The word “patent” only appears once in the Democratic Platform. It appears briefly in the section on healthcare: “For too long, prescription drug companies have gamed the system to justify their price increases by any means available. Democrats will crack down on anti-competitive efforts to manipulate the patent system or collude on prices.” Interestingly, a new FTC report found that patents were not the main reason for high drug prices; it was the middlemen buyers of pharmaceuticals, called PBMs, that run up prices. The word “intellectual property” only appears three times, all of which are in reference to preventing the “theft of intellectual property,” specifically from China. It is clear that the platform does not view IP as deserving its own protection as a special class; it is bundled with other global economic competition concerns, such as currency manipulation.

Weekly Novelties

• The rise of animated GUI design patents, especially for animated icons (PatentlyO)

• Congressman Massie was “appalled” by efforts to limit the ITC’s power over infringers (IP Watchdog)

• Researchers claimed that USPTO’s software has mistakenly resulted in longer-than-expected patent life extensions (Stat News)

• Samsung got sued over mobile device charging (Bloomberg Law)

• An analysis found that carbon capture patents impact the insurance industry; patents can have downstream effects on surprising industries! (Carbon Herald)

• USPTO was affected by the Crowdstrike IT outage and is still dealing with the fallout (USPTO)

• XPeng is licensing patents to Volkswagen, opening up a new line of revenue and highlighting Volkswagen’s dependent on others, like Rivian, for EV innovation (IAM)

USPTO came out with more information on this topic just a few days ago. Specifically, there is now new guidance in the Federal Register as well as three new examples of inventions to help patent practitioners clarify patentable subject matter. But what does it mean?

This week on Nonobvious, we are going to briefly digest this USPTO guidance on AI subject matter patents and what they mean for practitioners.

USPTO AI Guidance Analysis

USPTO provides several non-limiting examples of non-abstract ideas in its guidance, including:

1. An ASIC for a neural network

2. A system for analyzing livestock health that uses radios and machine learning and specifies which variables and data are used

3. A treatment method for treating the disease NAS-3

There are also references to a number of examples from case law, including a specific RFID encoding. None of these examples should be particularly surprising to practitioners who have spent a lot of time with AI. First, these are all applications rather than artificial intelligence methods on their own. One is a hardware invention built for artificial intelligence; the next is a system (not a method!) that combines radios. Also of note, the livestock invention specifies the specific variables and data relating, which relate to movement. And the last is a treatment method, which is common in the life sciences.1 The guidance emphasizes the importance of proving a true improvement, such as an improvement to a computer in a technical field. The guidance is firmly set in the Alice/Mayo two-step analysis. Readers of Nonobvious will already know a number of these tips from last year’s AI patent guide.

Yet, the guidance is unable to avoid some of the ambiguity of the abstract idea concept. Examiners are also supposed to evaluate whether the method represents an abstract idea or mental process. But the well-known complaint with this aspect is that mathematics and computation are a key ingredient in any computer invention; arguably, all code is just executed bytecode executed mathematically by a machine. So there remains a question as to where the line is in “abstraction.” USPTO tries to help by providing three examples.

They are:

• Example 47: An artificial neural network that detects anomalies. The first, which is eligible, recites an ASIC with specific limitations to the neural network in question. The second claim, which is a method for training and using the neural network, is ineligible because it recites a very abstract idea while not connecting it to the practical application. The third claim, however, is valid because it applies the neural network to the specific application.

• Example 48: An algorithm for analyzing speech. The first claim is ineligible as abstract and just math. The second claim, which is dependent on the first claim, is eligible because it specifically improves speech separation and is not “directed to” the abstract idea exception. And the third claim, a computer readable medium, integrates the second claim into a practical application.

• Example 49: A method for treating fibrosis using a machine learning method that applies several known treatments. There are two broadly similar claims, but the first one is ineligible as abstract for failing to connect to the practical application while the second one is. The main difference is that the second claim specifies a specific treatment rather than merely suggesting, as the first claim does, an “appropriate treatment.”

Subscribe now

I find Claim 3 of Example 47 the most interesting. Here it is:

[Claim 3] A method of using an artificial neural network (ANN) to detect malicious network packets comprising:

(a)  training, by a computer, the ANN based on input data and a selected training algorithm to generate a trained ANN, wherein the selected training algorithm includes a backpropagation algorithm and a gradient descent algorithm;

(b)  detecting one or more anomalies in network traffic using the trained ANN;

(c)  determining at least one detected anomaly is associated with one or more malicious network packets;

(d) detecting a source address associated with the one or more malicious network packets in real time;

(e) dropping the one or more malicious network packets in real time; and (f) blocking future traffic from the source address.

This is an extremely limiting claim. It does not require you to describe the training data (interestingly!) but does require specifying a particular machine learning algorithm. it also describes the specific actions the neural network and system as a whole are supposed to take based on the neural network’s predictions. This suggests to practitioners to try and tie neural networks to specific, but extremely broad, applications (you could plausibly define gradient descent to capture nearly all machine learning methods today, including transformers)2 and to tie it to deterministic outcomes, like another algorithm or a physical device.

Despite these features, this claim has much in common with some of the example ineligible claims. For example, Claim 1 of Example 48 has no practical example but does go into great detail into the specific model, describing the use of Fourier transforms. Similarly Claim 2 of Example 48 seems to me to be just as abstract as Claim 1, so I am not quite sure why USPTO feels this is more directed towards an improvement. USPTO’s Alice/Mayo analysis emphasizes aspects from brief specification that describe how these additional steps are an improvement over using deep neural networks with Fourier transforms and specifically attach these improvements to the specific application, so perhaps there is a lesson in using the specification to help explain the non-abstractness of the claim. Similarly, Claim 1 of Example 49 describes a machine learning method that is applied to the life sciences. In fact, although it is less specific as to the machine learning method, it is more specific as to the types of data it uses (SNPs), but completely unspecific in the particular action taken (dropping offending packets vs applying an “appropriate treatment”).

I am a known critic of the vagueness of the abstraction rules. But my takeaways are these:

• If you don’t have a practical application for your machine learning method, don’t bother

• There is a sort of menu of abstraction: the machine learning method, the data used, the actions taken, and you need to pick from this menu but don’t necessarily need to have every one in your claims

• The specification is important, and needs to tie your methods back to improvements and practical applications

• Try to not do pure algorithm claims in the form of methods, though it is possible to get those granted

Meet the Author

Are you at NAPP today? Come see Evan! We’re exhibiting Edge and he’d love to see you and show you the best patent software in the business. Even if you can’t come, reach out and he’ll give you a demo.

Try Edge

Weekly Novelties

• Dennis Crouch also put out analysis of the USPTO guidance on PatentlyO (PatentlyO)

• Some have accused pharmaceutical companies of watering down a bill to fight patent thickets (State News)

• One analyst claims that the patent cliff’s biggest losers will be Merck, Bristol Myers Squibb, and Amgen (Fierce Pharma)

• The EFF claimed victory in efforts to limit a proposed rule that would have limited discretionary denials of IPRs (EFF)

• In SIPCO, LLC v. Jasco Products Company, LLC another fun case about a small typo with big consequences (Law360)

• Sanofi and Regeneron convinced the new UPC court to overturn a key Amgen patent, marking the first major invalidation of the UPC (Juve)

• The UK announced that it is beginning a six-month pilot for a new digital patent system; the first application filed under this new system was done by the firm Murgitroyd (Law360)

The legal industry is a trillion-dollar, competitive industry, so if AI can introduce real efficiencies it is coming, like it or not. However, there is an interesting economics principle that suggests that it might not be all doom and gloom if AI can bring efficiency to the legal profession. In fact, according to this principle, lower billable hours per matter may mean more billable hours overall. How? The answer has to do with highways and traffic. Enter Jevon’s Paradox.

Billable Hours, Price Elasticity, and Traffic

Have you ever wondered why the traffic gets worse whenever they expand a freeway? It turns out there is a reason, and it comes down to economics.

The idea comes from English economist William Stanley Jevons. In 1865, he noticed that increased efficiency in the coal industry lead to an increase in aggregate coal consumption. Standard convention at the time was that technological progress would reduce overall consumption of all resources; he argued the opposite. It turns out that he was right. This has been observed in a number of different areas of the economy over time. In the 1980s, for example, Daniel Khazzoom and Leonard Brookes predicted that increased energy efficiency would mean that overall energy consumption would increase; this turned out to be true. This has also been observed with crops. And famously, it has been observed in traffic patterns.1

The reason is what is called the “rebound effect.” Essentially, when you become more efficient in using a resource, the amount of resource used per task decreases, reducing the demand for that resource overall. But sometimes this unlocks “hidden” demand—people who wanted to use the resources but couldn’t justify the cost2 at the previous levels of efficiency. When the “price elasticity” is high—meaning that if you cut the cost of something by $1, you get more than $1 of demand—you end up unlocking more hidden demand than you eliminated through efficiency. That is the beating heart of Jevon’s paradox. You can see the connection to legal work. While the cost of a service may go down, if there is enough “hidden demand” for legal work, the overall spending will go up because there will be more matters to handle, even if the cost per matter decreases due to efficiency.

We have seen examples of this in the legal world in the past decade. Two from corporate work are Stripe Atlas and Clerky. Stripe Atlas handles incorporation matters for startups in a cookie-cutter manner for a flat fee. Previously, startups would be charged $10,000-20,000 for formation matters. Now, they can use Atlas and get access to Goodwin-drafted documents alongside process automation and deals on startup services, like bank accounts from Mercury and credit cards from Brex. Similarly, Clerky generates certain types of standard form contract drafted by Orrick, like NDAs and SAFEs. For a one-time flat fee, startups can get access to these documents and use them as many times as they like. These standard documents would often cost thousands of dollars from a good firm. And they weren’t even that profitable; they were tedious jobs that lawyers disliked doing and distracted from higher-order matters.

Stripe Atlas and Clerky have had an effect on the legal profession’s pricing power for these services. White shoe firms will now offer a $5,000 package to get startups on board that is similar to Stripe Atlas. Similarly, prices for standard agreements have gone down. Edge is represented by well-heeled counsel and was advised to not bother using a form offer letter from them. But business overall is up. Startup formation is at a record high, which means that although the fees are lower, the quantity makes up for it. Similarly, while companies are less likely to get bled dry for simple contracts, they are more likely to find value from their lawyers and feel more comfortable asking for complex work because the legal budget has not been consumed by pedestrian work. Certainly, that has been our experience. And in patent law, an initial draft is often a flat fee, but that has not meant less profitability overall for the field. Patents are higher-quality and more complex than ever.

Knowing that AI is coming, lawyers can respond by understanding how to procure well, whether outside counsel or in-house. They can be ahead of the competitive curve by being early adopters of tools that work for them and treating obstacles, like privacy and security, as hurdles to overcome instead of blockers. They can also think about how to use AI to generate efficiencies in their practice that are cost centers today, like in training and staffing for operations. And lawyers can retool their offerings for clients. With the efficiencies of AI, how can you improve the quality of what you are offering to create a higher-quality product and, therefore, a higher billable hour? Are there things that are expensive today that you can offer for cheap to bring clients in and upsell? Or what matters that take time away from more profitable matters can be done quickly to reduce drag on the practice overall? And most importantly, thanks to increased efficiency, what can you offer today that before was impossible, thus creating new legal products altogether?

Try Edge

Lawyers have a choice in how they react to this shift, but it is inevitable. AI introduces efficiencies that will allow law firms to compete over business. Thanks to competition, AI adoption is coming anyways. The question to ask is not whether AI will be bad but how to make it good. And it should be noted that the law is a competitive space and it always has been. Increased competition and efficiency has not stopped the profitability of the legal industry so far. In fact, AmLaw 100 revenue grew this year to a record high.

Meet the Author

Are you coming to the NAPP Annual Meeting in Minneapolis next week? Stop on by! Evan will be coming as well. Stop by the Edge booth. Reach out to schedule a meeting. And if you want to sign up for Edge, click the button below to get started.

Sign Up for Edge Now!

Weekly Novelties

• Dennis Crouch shares some views on what Loper and the death of Chevron means for USPTO. We’ll post something on Nonobvious soon as well on this topic (PatentlyO)

• The FTC backed USPTO’s proposed terminal disclaimer policy; others say it would be bad for patents and argue it exceeds the agency’s authority (IP Watchdog)

• Dexcom’s first lawsuit against Abbott at the new Europe UPC failed with a court finding Dexcom’s glucose monitoring patent was invalid in its entirety (JUVE Patent)

• A WIPO report found that Chinese inventors have filed for 6x more AI patents than American ones, more than anyone else in the world; however, American patents are cited more (VOA News)

• The DOJ appealed a decision by a decision that Gilead did not infringe on three CDC patents, and that furthermore the patents were obvious; evidently, drug patents are acceptable when they are the government’s (Fierce Pharma)

• Commissioner Vidal directed USPTO to review PTAB decisions rejecting challenges to an airsoft gun patent, saying that the panel improperly relied on the personal anecdotes of expert witnesses in determine whether there was a motivation to combine (Bloomberg Law)

• The Niskanen Center, a prominent think tank, came out against the USPTO terminal disclaimer proposal (Niskanen Center)

• The FTC’s battle against the Orange Book entered a new phase where the FTC challenged several medical device patents it alleges are improperly listed (Citeline)

• USPTO and UKIPO announced a collaboration effort around SEPs (Akin Gump)

To celebrate Independence Day, let’s look at a patent, specifically US 2,063,601 A, which was filed in 1934 and granted in 1936. That patent is for the sparkler, which is today used across celebrations.

Sparklers were invented by August Hummel. Although sparklers existed before, they were not flexible like Hummel’s invention; they also would often deteriorate while in storage, rendering them useless as a widespread commercial product. Thanks to Hummel, we now have handheld sparklers. Wonderful!

Weekly Novelties

Although there is not much news this week, I do want to point your attention to the newest DABUS decision last week, this time in Germany (here is an article in Juve). The court held that while DABUS could not be an inventor, the person who directed DABUS could be.

AI is only becoming more important to the inventing process. There will need to be more work to address how patent inventorship should reflect this change in the scientific process.

But once you’ve decided to buy, what is the best way to do it? You want to make sure that it is not too strict, but has guardrails to prevent errors; that you are gathering the appropriate level of information; that you aren’t going too slow and lose out in the competitive landscape; and more. This week in Nonobvious, we cover some tips to be bring your procurement process to the next level.

This is part of a series to help lawyers get better at buying software.

To see Part 1 of our procurement guide for outside counsel, look here.

To see our guide for procurement for in-house counsel, look here.

Master Your SaaS

Types of products

First, let’s talk about the types of products you might buy. When you take the time to decide what you truly need, often you may find that there are several functions that need to be solved to achieve a certain objective. There are two main approaches here:

1. Best of breed: Here, you buy a different product for each core function that is the best at what it does and stitch everything together. The positive of this approach is that you get the best solution for everything and have more leverage over a vendor because you are less dependent on them. The downside is that the integrations may not exist or require a lot of work.

2. Full-service: Here, you search for a vendor who can do it all. The benefit of this approach is that there are often benefits to integrated tools and the procurement process is often simpler because you are only dealing with 1-2 entities. The downside is that the vendor becomes more essential to your business and that full-service solutions often do not integrate well with other solutions, so if they don’t have everything you need you could be stuck.

One subtlety of the best of breed approach is that it is more amenable to startups. This is a huge positive. Startups are often more innovative and attentive to the needs of their customers. While it is easy to be risk-averse and look for a well-known name, I cannot encourage you enough to take startups seriously as vendors. Especially in AI, where almost everyone is a startup, this will help you access great technology faster.

The best of breed approach is best for work that only affects a single team or person, where integrations are not important, and where the purchase decisions are not massive. It is also almost always the best for products that are using a new technology because the incumbents will often do a poor job designing them. The full-service approach is best when integrations between different aspects of a job are important or in very mature sectors where a large company can not only meet your needs but plausibly do a good job with them at a volume discount.

Who decides to buy?

Next, let’s talk about who controls the purse strings.

Law firms have historically almost always engaged in top-down purchase decisions where feature checklists rule the process and consensus among all the partners is required.

Checklists are almost always a mistake to follow strictly; it is better to determine which features are important and which are nice to have. The real world has trade-offs, so your procurement process should too.

Particularly for tools that support individual contributor work—for example, an AI drafting tool—you might want to adopt a “prosumer” model. Here, individual lawyers get their own budgets to spend on tools that support their work instead of going through a managing partner or a committee. There is no obligation to ensure the tools are the same or even do the same work; they just need to meet certain minimum requirements. This allows individual team members to solve their needs. It is also a decentralized way to figure out the best tool. If one tool grows rapidly in your organization, or is clearly superior in the metrics you track, you can then reach out for an enterprise license and go top-down.

The prosumer model is particularly amenable to law firms because of the control individual partners have over their teams. You might even consider allowing associates to be prosumers since they can then buy tools that they need to do their work; a complaint I often hear from associates is that tools that would support their work isn’t prioritized because partners make all of the purchasing decisions but often do very different work. However, partners often don’t have the time to get too deeply involved. This can help resolve that tension. It would be a big shift because associates have historically been given no control over their work or budget. But in the software world, junior engineers are given a (small) amount of control over how they work and this has increased productivity significantly.

Subscribe now

In fact, this approach has become very popular in Silicon Valley and has been good both for buyers both by making teams more nimble and by helping support a more vibrant vendor ecosystem. This approach doesn’t work for everything, but it can be more flexible than you might think. For example, when choosing between Teams and Slack, you might think this has to be a top-down decision, but you could instead push it down to the team level.

How to improve procurement at a firm-wide level

For the managing partners out there, here are some tips for improving your procurement posture on an organizational level:

1. Let partners own their budgets. Many partners do not own the budget for their team; rather, purchasing decisions are made on a partnership basis. You should fight for your own budget that you can control; don’t surrender to the law librarian. Importantly, make sure you also get a pilot budget that you control. Many software companies will not do a trial without a paid pilot to show the seriousness of the buyer. The amount will depend on the scope and scale of the pilot, but it could be anywhere from $10,000 for 4 weeks to $100,000+ for several months per paid pilot. This will also allow you to move faster.

2. Have teams practice ROI calculations. An important part of this is getting used to ROI calculations. Often, lawyers only consider the cost of software and think of it as money out of their pocket. While that is true (and it’s great you’re thinking like an owner, which you likely are), there are also benefits. It may be time saved so you can focus on higher-order matters for your clients, costs saved, or something else. But once you start to really look at the benefits of something, not just the costs, you will start to have a better sense of what is worth buying. This will likely mean your firm starts buying more software, but it will also help you avoid buying duds.

3. Get to know your law librarian and IT. In a typical procurement process, there are two well-known roadblocks to deals. The first is legal. And as a law firm, well, you can get through that barrier quicker, but this does require engagement (see next tip!). But the other is IT. First, maintain strong relationships so they will prioritize your needs. But more importantly, know the core requirements and have a strong grip on the process. Are there core requirements, like SSO or security certifications like SOC2 or ISO 27001? What is the normal timeline? As for the law librarian, at many law firms, the librarian actually owns the procurement process. If this is the case, you should try to change that. But the law librarian will always be an important partner in a purchasing process. Get to know them. They can provide experience and give context that you may have not considered, especially with your busy job doing legal work.

4. Own legal compliance as much as you can. It is quite common for companies to rely on outside consultants in generating compliance requirements for a purchasing organization. Law firms are no different, but it can get a little absurd. I have heard from a world expert in privacy law who was told by his compliance team that he couldn’t do something because it created a privacy issue. (You can imagine how well that went over with him.) The best way to avoid this issue is to engage in the setting of these policies before you need to buy something. This may also involve speaking with your malpractice insurance provider far in advance to know what is in-bounds, out-of-bounds, and what the process looks like for gray areas. Just know that malpractice more has to do with how you conduct your firm more so than what software you buy.

5. Set firm-wide priorities with wide latitude. Big decisions on things like budget, priorities, and more are often set by big decision makers in an organization on an annual and quarterly basis. Consider doing this on a partnership level. However, there needs to be significant latitude. Different teams and groups have different needs, and it is important for firms to accommodate this so that they can be effective. Often this will mean dividing budgets into a general firm-wide budget used to achieve firm-wide priorities while having some of the budget set aside for partners to make their own decisions.

Procurement processes are a somewhat general process, but sometimes there can be elements that are unique to a particular industry. For example, the pharmaceutical industry will often provide blind tests to vendors to see if their product really works on questions that they do not know the answer to because of the complexity of the field. In the case of law, I do not believe that there is anything as unique as the pharmaceutical industry, but I do believe that there is a need to focus on two common aspects more than other fields do.

The first is accuracy. While many areas of the law require creativity, lawyers have a duty of competence and can face severe consequences for mistakes—though most of the fears surround sanctions, a much more common problem is the fact that many legal actions are immutable. This can be handled by testing or asking for references. You must think about what the acceptable bounds are for accuracy—perfect is often impossible—and which categories require more accuracy. Let’s take the example of writing a brief. On the one hand, it is an art, so there is no “right” way—100% accuracy is nonsensical. On the other hand, there may be some things where accuracy is very important. For example, if a brief cites a case, the citation should exist and be correct 100% of the time. The summary of the case is up to opinion but less subjective, so determine what “accuracy” means and how you test it.

The second is confidentiality. Lawyers have the benefit of privilege and a duty of confidentiality. While this does not mean that lawyers have a duty to use super-measures, but it does mean that lawyers have an obligation to review the confidentiality of every vendor they use. State bars have been addressing this issue for over a decade now and you should simply follow this guidance. On a practical level, most of these issues can be handled by reviewing policies and asking for common cybersecurity certifications. If someone is SOC2 or ISO 27001 certified and has a reasonable privacy policy this should be enough for 95%+ of cases.

On one final note, you may consider design partnerships. This is where a buyer partners with a vendor to provide them higher levels of support and feedback, and sometimes data, to help the development of a product. This will sometimes accompany an investment in the vendor. In exchange, the buyer will often get access to the product early, have their needs prioritized on the product roadmap, and get preferred pricing.

Prior Art

Previously on Edge, we covered the 2028 patent cliff. Nearly $100 billion of Pharma revenue is at risk from patent expiries. For some companies, nearly 1/3 of revenues could come under threat.

This week, Biospace has an interesting article covering the different approaches taken by three Big Pharma companies: Pfizer, Johnson and Johnson, and Bristol Myers Squibb.

Weekly Novelties

• The UPC officially opened its Milan division (Law360)

• In a landmark en banc Federal Circuit ruling, the bar was raised for the second prong of the Alice test with a requirement that the claims be “explicitly linked” to the specification (World IP Review)

• Starbucks was sued over a French Press coffee maker over patents held by some of its partners (Bloomberg Law)

• An Apple patent showed a radically different iPhone design with a removable back panel. Though it could be meant to more easily comply with “right to repair” laws, it also brings to mind the mid-2010s modular Android phones (Forbes)

• Although Microsoft canceled its Xbox cloud, the patents still come out (The Verge)

• The UPC’s first substantial rulings will be coming out in the coming weeks; this analysis goes into what they could mean (Juve)

• In a valuable venue ruling for defendants, the Federal Circuit held that Haptics, a Texas company, must sue Apple in California for convenience (Patently-O)

Try the platform

Industry-Standard Accreditation

The SOC 2 audit is one of the highest recognized standards of information security compliance in the world. It was developed by the American Institute of CPAs (AICPA) to allow a third-party auditor to validate a service company’s internal controls with respect to information security.

We obtained our audited SOC 2 Report by partnering with Johanson Group who respectively review our internal controls including policies, procedures, and infrastructure regarding data security, firewall configurations, change management, logical access, backup management, business continuity and disaster recovery, security incident response, and other critical areas of our business.

Schedule a demo

Our report was unqualified, meaning that there were no exceptions identified.

On top of the standard SOC 2 requirements, we added a custom control verifying that we do not train on customer data without permission. That is part of our report. We did not need to add it; in fact, it added an extra hoop for us to jump through. But it was that important for our customers to know we were keeping our promise.

Why does this matter?

Bars all over the country allow for the use of cloud software and they are increasingly coming down on the side of allowing AI software as long as appropriate precautions are taken. In fact, USPTO has even come out to say that the use of AI tools is allowed and not required to be disclosed!

Yet, all of these permissions still have a duty of care. You need to make sure your client’s data is private and that it is not being trained on to prevent regurgitation.

First, by achieving SOC 2, we remove much of the uncertainty around working with us. Not only do you know that an auditor verified our cybersecurity standards to the highest standards, you know what those standards are. This should give greater assurance to your IT teams, compliance teams, and even your insurance.

Our security

Thanks to a company-wide effort here at Edge, and with the help of our trusted partner Johanson Group, we successfully achieved SOC 2 compliance and received an Auditor’s Report, which we are happy to share with you to prove to you that our policies, procedures, and infrastructure meet or exceed the SOC 2 criteria. In fact, by partnering with Johanson Group, we can confidently say we go above and beyond the minimum requirements for SOC 2 by integrating our critical infrastructure to monitor compliance to the SOC 2 framework 24/7/365, not just during the audit window.

We believe the relationship with our customers must be built on trust. The successful completion of our SOC 2 Report is one of many ways that we have planned to earn and retain that trust.

SOC 2 is just one aspect of our growing security program. We are committed to continually improving our information security program and retaining an annual SOC 2 audit to ensure we keep supporting our customers’ needs.

If you want to try what is now the most private and secure patent platform in the market, click below.

Try Edge today

Procurement is the art of buying resources. Traditionally, doing legal work just required a few people in an office, along with basic software like Microsoft Word and legal research tools like Lexis or Westlaw. Buying new software wasn't something many law firms needed to think about often. That is the reason procurement for lawyers is typically done with no real process.

This means that law firms are typically bad buyers. Being a good buyer means making decisions quickly and fairly. A lack of good buyers hurts the industry overall because the ecosystem cannot sustain a critical mass of good tools without sufficient buyers. This creates a negative cycle where lawyers are deprived of the tools they need.

But times are changing fast, especially with the rise of AI technologies that can help lawyers do their jobs better. There is more demand than ever before. To help this budding industry take hold lawyers are going to need to learn how to run a tight procurement process.

This week in Nonobvious, we are going to dive in to what procurement is and why you should be good at it. In Part 2 next week, we will dive into tips for being good at procurement as an outside counsel attorney.

Subscribe now

This is part of our series on procurement for lawyers. To learn more about this for in-house counsel, look here.

To see Part 2 of this series going over tips to sharpen your procurement process, look here.

The First Step In Fixing Your Problem Is Knowing You Have One

Buying new software is critical for productivity. It can give your firm an edge when competing against others and offer a better product more cheaply for your client. Because lawyers are not used to buying software, they are often inefficient at buying software. This hurts the lawyers, their bottom line, and the entire ecosystem.

At law firms, procurement is typically not handled the way you would expect a good process to run.

Procurement typically has a certain set of steps:

1. Identify Your Needs: First, figure out what you really need help with in your practice. This is usually done by engaging with the stakeholders

2. Research: Look up different software options and list their pros and cons.

3. Talk to Vendors: Set up meetings with companies that make the software to learn more.

4. Compare Options: Sometimes you can do a trial or a test run (called a 'bake-off') to see which software works best.

5. Narrow It Down: You might want to have a second round of reviews if you have a few good options.

6. Check Everything: Make sure the software passes legal and IT checks.

7. Make a Decision: Choose the software and start using it.

There are a number of ways that this breaks down in practice.

First, the key people are often not empowered in the decision making process. Associates who would be using tools have no buying power and typically are not even allowed to suggest tools. At boutique firms, buying is usually ad hoc and subject to the interests of managing partners. Senior counsel and of counsel cannot bring in the tools they like and are used to. At full-service firms buying is usually done by the law librarian with input from IT, but this is often done without deep engagement of the lawyers, who are usually totally unaware of what their buying process even looks like.

Second, lawyers are very busy people. They often do not have the time to compare options, research tools, try them out, and do training. Most vendors do not realize the amount of training they have to do. Certainly, they do not have the time to learn how a complex procurement process works. Once, I met a leading privacy scholar who had to argue with his IT team about what a privacy law meant for his firm’s procurement process—and he lost that fight! This is what happens when you don’t have time to engage with the process.

And for those reasons, lawyers are often very risk-averse. They don’t know how to check whether a company could create a privacy problem or which resources to turn to. They don't realize what a good length of time to review software is. This leads to lawyers being afraid to try out software from newer companies even though they often have some really cool stuff that the big guys don't have yet.

Most lawyers will say this is of necessity. After all, with billable hours, time is literally money.

But it isn’t actually that way. Even if you believe it is.

Buying software isn’t just about spending money—it’s about investing in tools that can make your work better and your clients happier. By learning how to choose and use the right software, you can make a big difference in your practice. Yes, it requires a time investment, but it can be done efficiently and is an investment in your practice.

Next week we will talk about some “do”s and “don’t”s in buying software and what outside counsel specifically should keep in mind.

Check in next week for Part 2!

Prior Art

In March, we covered the 2023 global drop in patent filings. This is primarily driven by a drop in American patent filings. But some places have not fallen. China is one such jurisdiction that did see growth last year. It is now the largest jurisdiction for patent filings.

This quarter’s special report in The Economist explored China’s rise as a superpower. Entertainingly, the story starts with a wall of 192 patents at the Chinese Academy of Sciences for various seeds. It is worth a read to understand some of the metrics of Chinese growth, such as scientific publications. Research indicates that Chinese patent quality is improving. Perhaps it will continue further.

Try Edge, draft patents faster

Weekly Novelties

• Cert was filed in Chestek v. Vidal, a case in which the Federal Circuit held that USPTO was exempt from certain APA requirements for notice and comment periods. Administrative rules are a particularly interesting focus given the potential review of Chevron that is on the table. Five amicus briefs have already been filed (PatentlyO)

• Subject matter eligibility rejections are dramatically increasing. Notably, the AI group now includes a 101 rejection in 77% of cases even though AI applications are soaring overall (Voice of IP)

• A new analysis shows that the rise of the UPC is leading to a massive drop-off in patent cases in Germany as litigators chose a different venue (JUVE Patent)

• Impossible Foods, the original “vegan meat” company, invalidated 23 out of 24 claims in a key patent at the PTAB (Agfunder)

• GE Aerospace came to an agreement to sell their patent portfolio and licensing arm for $429 million to Dolby Labs, most notably including thousands of patents for video compression codecs (Bizjournals)

• Mimir, an NPE which purchased IP from SK Hynix, has sued Micron. Proxy warfare is growing, particularly in the semiconductor industry (KED Global)

• The FTC endorsed a USPTO proposed rule that would expand the circumstances in which settlements must be published (Bloomberg Law)

• IBM settled its dispute with Rakuten, claiming that it was infringing on patents related to its pioneering 1997 ecommerce platform, Prodigy; terms were not disclosed (DigitalCommerce360)

LKQ overturned the Rosen-Durling1 standard for design patents, which required that the primary reference be “basically the same” while the secondary reference must be “so related” that the related features practically suggest application. If this sounds a little like “obvious to try” to you, you’re right—the Federal Circuit explicitly said that it was trying to bring design patent nonobviousness in line with KSR (and, in a retro twist, an older case, Smith v. Whitman Saddle Co., 148 U.S. 674 (1893)). This week in Nonobvious, we’ll go over what this might mean for design patents.

Designs On You(r Patents)

Design patents are not a large enterprise. According to USPTO statistics through 2020, there were 47,838 design patent applications in 2020, representing a mere 7.2% of all applications. However, they are growing in popularity. Design patents grew at an annualized rate of 5.1% over the past decade, which was much higher than the 1.6% annualized growth rate of patent applications overall. They are also much easier to get. Design patents have a 3 year grant rate of 97%—yes, you read that right—compared to 75% for all patents. USPTO statistics also show that nonobviousness rejections are extremely low for design patents, a mere 7% of all office action rejections.

That said, design patents may be used for a variety of purposes and are important for those who use them. For unique products, they can protect shapes, textures, and more. It can even cover the shape of cereal. In addition to covering designs for their uniqueness, design patents can also be used to protect razor/razorblade business models. For example, Keurig used to have a design patent for its coffee pods, effectively blocking anyone from making unlicensed pods by controlling the compatible shape with their device.2 They are often used in categories where trade dress and copyright are inapplicable, so protection for unique design elements may be otherwise totally unavailable.

The Federal Circuit did not give a great amount of guidance, unfortunately, for how to apply this ruling. It did say that the “primary reference will likely be the closest prior art” by visual similarity and that the “primary reference will typically be in the same field of endeavor” as the claimed ornamental art unless it is analogous art. This is not much to go by. Professor Dennis Crouch at PatentlyO points out that the Federal Circuit cites In re Jennings, 182 F.2d 207, 208 (CCPA 1950), which says that the prior art must be considered at a whole, which is at least something to work with. Yet, much was left unanswered. For example, the Federal Circuit declined to determine whether the two-pronged analogous art test for utility patents from In re Bigio, 381 F.3d 1320 (Fed. Cir. 2004)—where the art must be (1) from the same field of endeavor as the claimed invention or (2) reasonably pertinent to the particular problem addressed by the inventor, regardless of the field—applies to design patents.

Due to that lack of guidance, USPTO has not been able to give much direction, either. After all, the Federal Circuit called the old approach “too rigid” and for now called for a “case-by-case approach.” USPTO released a memo the next day, but by and large it emphasized that examiners should be looking for some documented reason why an ordinary designer would add features from a secondary reference to a primary reference. In other words, until there is more clarity from the courts, it seems like USPTO will not be making too many changes to how it grants design patents (yet).

But of course, there is still much subjectivity with designs. And the court went out of its way to say that it was deriving from KSR because the language of 103 does not distinguish between utility and design patents. So while practitioners for a long time thought KSR did not even apply to design patents, there may be more holdings in the near future that unsettle design patents even further, especially given how many questions were explicitly left for later.

Prior Art

Last edition, we covered the blow-up with OpenAI and Scarlett Johansson. As we observed, there are inconvenient facts for Johansson, but overall it is a strong case, and if it is as she describes it is nearly a slam dunk.

Since then, Johansson has turned into an unlikely avatar for issues surrounding AI. In addition to becoming the symbol of a proposed federal right of publicity, her case has also become the example a lobbying group used to push for a federal law outlawing deepfakes. She also might be asked to testify in front of Congress.

Despite all this, she has not yet filed suit.

Weekly Novelties

• This week, the FTC notched a rare win in its fight against drug patents. A judge ruled that five Teva inhaler drugs were improperly listed in the Orange Book (Fierce Pharma)

• As the US negotiates a renewal of science and technology pact with China, lawmakers are asking for Chinese patent data to understand if US federal funding fueled Chinese patents, especially as concerns over IP theft reached a boiling point (Reuters)

• A survey of IP lawyers found that they are seeing increased adoption of AI, but some concerns about accuracy remain (FT)

• Guardant sued Tempus over several DNA diagnostic patents. The liquid biopsy market remains hot and patent fights abound (Bloomberg Law)

• Samsung sued Oura preemptively to avoid potential issues over its patents. Oura aggressively asserts its rights and even filed for an ITC investigation against competitors like Ultrahuman (VOI)

• Dolby acquired GM Licensing and over 5000 patents for $429 million (IAM)

• Adobe updated its terms of service, saying it would no longer train on customer data for its models. Once again, non-fair use considerations, like angry customers, remain more important for determining the bounds of AI training rights (Bloomberg Law)

• A patent fight is brewing between ARM and Qualcomm, and it could upend the fight for AI by holding up some of the chips used to build those systems (CNBC)

Traditionally, a disclosure is the result of a structured intake: the inventor fills information into a form, that is then manually turned into a disclosure document after significant practitioner time is spent interviewing the inventor. The result is manual and wastes everyone’s time.

Edge solves this problem today with its Disclosure function, which is interactive and focused on a disclosure form.

Today, our Disclosure functionality is much more interactive and flexible. And most importantly, it lets you get up and running much faster than before.

Try Edge

Disclosure’s Got a New Groove

Now, starting a project with Edge is just the press of a single button, “New Project.” You will land in a page that provides several options for creating and managing Disclosures.

You are presented with four options:

1. Start Disclosure with a chat: This is an interactive, updated disclosure experience. Just upload what you have and Edge will create a disclosure for you and then ask follow up questions in a single thread.

2. Update an existing Disclosure: This allows practitioners to upload updates to a disclosure, such as a transcript or notes, and update the disclosure.

3. Upload completed Disclosure form: For high volume clients, they often have a standard disclosure form that has been completed per their standard processes. This functionality allows you to upload a completed such disclosure without any modifications or changes.

4. Start Disclosure with a form: This is our classic interactive disclosure form, complete with our interactive disclosure question generator.

Try Edge today

In the example below, we take a document written in German, summarize its contents in English, and answer a clarifying question.

The Assistant can process documents in DOCX and PDF format, powerpoint PPTX presentations, and JPEG and PNF format images, as well as additional context provided by explicit instruction.

Updating a disclosure works similarly; the only difference is that instead of starting from scratch, the Assistant starts from your existing disclosure and updates it.

We have many more features coming to improve Disclosure. Want to see more? Reach out for a demo.

Write better patents with Edge

Subscribe now

In an interview with the Verge, Haynes Boone’s Purvi Patel Albers describes this as a slam dunk for Johansson. In the article, he is quoted as saying “The question we need to ask is why on Earth did OpenAI do this?…[Johansson] is a known advocate for protecting her rights, so she’s not going to shy away from going against them.” That said, the case is not so cut and dry. This week, we’re going to review some of the rules behind right of publicity to get readers up to speed. We’re also going to put this into context of the broader IP rights regarding AI; this is the opening salvo of non-copyright battles, and it will be more interesting than the question of fair use

.

Also, before you continue, I was interviewed by about patent law and AI. Listen here:

AccelPro | Intellectual Property Law

On Patent Prep and Pros Assisted by Technology

Listen on Apple Podcasts, Spotify and YouTube Welcome to AccelPro IP Law, where we provide expert interviews and coaching to accelerate your professional development. Today we’re featuring a conversation with Evan Zimmerman, CEO and Co-founder of Edge…

Listen now

2 years ago

Voice Misappropriation and GPT-4

Voice misappropriation is a relatively new area of the right of publicity. The right of publicity, which is a bit of a strange beast that is a mix of a tort and a property right, exists on a state-by-state basis. The controlling cases are Midler v. Ford Motor Co., 849 F.2d 460 (9th Cir. 1988) and Waits v. Frito-Lay, Inc., 978 F.2d 1093 (1992). Specifically, the worst is “the deliberate misappropriation of (1) a voice, that is (2) distinctive and (3) widely known” (Waits at 1100). Both cases involved companies that hired voice actors to imitate famous musicians endorsing products using their actual songs; the tort is when the voice is “deliberately imitated in order to sell a product” (Midler at 463). It is no accident that both holdings involved singers because voice misappropriation requires not only that the voice holders are famous but also that the voice is “a sufficient amount of [their] identity” (Midler at 462).

More generally, the right of publicity generally has some conflict with the First Amendment. For instance, in Comedy III Productions v. Saderup, 25 Cal.4th 387 (Cal. 2001), the court mentioned that the transformative nature of the work is important when “the value of the work does not primarily derive from the fame.” However, as in Carson v. Here’s Johnny Portable Toilets, Inc., 698 F.2d 831 (6th Cir. 1983), the more commercial the likeness, the more the right of publicity comes into play. Because the right of publicity is typically state-by-state, there can also be significant variation in rights due to legislation. For example, in Tennessee (the home of Elvis), publicity rights last post-mortem and are very strong. Currently, there is no federal publicity right, though there has been renewed interest in such a concept.

Johansson’s accounting of the facts depicts what at first glance is a textbook set of facts for a right of publicity. Johansson’s role in Her is one of her most famous, and in it she is specifically recognized for the use of her voice. Johansson claims that she was approached by Altman to voice a new GPT-4 voice, which she rejected. Months later, her friends and family observed that the “Sky” voice sounded like her. She also claims that there was a second attempt to connect with her agent to reconsider mere days before the model was released. So far, it is fair to take her at her word on the facts, at least, since so far OpenAI has only disputed the legal conclusions.

There are, however, some inconvenient facts for Johansson that she does not mention. First, the tort requires that a “deliberate imitation” is used; the right of publicity is not like trademark law with a strict likelihood of confusion test. The most difficult fact for Johansson is that the Sky voice was actually launched in September 2023. According to OpenAI’s statement, they had cast the voice actress and finalized months before even approaching Johansson; indeed, records reviewed by the Washington Post show that the voice actress was hired several months before OpenAI even approached Johansson. The voice actress herself claims that no one had compared her to Johansson before. But in both cases, the sound-alikes were hired to get around a rejection. The other main difference is that in Midler and Waits the voice actors were performing Midler and Waits songs and endorsing products. Here, the use of the voice is a generic voice assistant, and Johansson cannot lay claim to a technical functionality as her art, nor can she claim a mere concept. There are no references to Her in GPT-4, nor are they attempting to imply that Johansson has endorsed the product.

This is not dispositive; OpenAI could have hired the Sky voice actress to imitate Johansson just in case she said no, for example. But the point is that, while this is already embarrassing for OpenAI, actual liability would likely require Johansson to show not just that the voices were similar but that it was intended to imitate her.1 That is the kind of thing that would come out in discovery. But if there was any such intent, it would very likely be game over for OpenAI.

The damages would likely be in the single-digit millions. In Waits, the jury awarded only $375,000 in compensatory damages but $2 million in punitive damages. Here the damages would of course be higher given that it is no longer the 1990s—for context, Chris Pratt was reportedly paid $5 million to voice Mario in Super Mario Brothers—but this gives a sense of what we are talking about here. Nothing catastrophic for a company like OpenAI.

The Bigger Picture: IP Beyond Copyright in AI

The bigger picture is that this is the opening salvo in a broader, post-copyright conversation of intellectual property rights in an AI world.

Previously, we covered the idea of fair use and in the context of AI. Although the coverage in Nonobvious was intended to be fair and balanced, I have made no secret of the fact that my opinion is that a finding of copyright infringement would be bad law and worse policy. In fact, I wrote a (tongue-in-cheek) article on my personal blog when Sarah Silverman sued OpenAI to that effect. In it, I observed that while I thought Silverman’s copyright claims were bogus, that there will be some type of licensing arrangement. At the time, I wrote that “most likely, as in everything else relating to Hollywood, [licensing] will have to do with official branding and merch.” That is where I believe this is going.

Copyright and fair use is just one way that creatives monetize their work and fame. Likeness and publicity rights are another example. but there are many others. There are many others. Trademarks may be used to make clear that a company is not endorsing a product. Breach of contract may be used to prevent a company from scraping a website even if fair use would allow the use of the work itself. Newspapers and movie studios may have back catalogs that are not open to the public, or which are cleaner or better-labeled than the data available to the public. There are many types of intellectual property beyond copyright and fair use and they are all fair game. As we are seeing here, many of them are a better fit for AI than trying to shoehorn yet another piece of computing into copyright law. That is why newspaper companies are inking deals left and right; they have other sticks and interests.

Weekly Novelties

• Several former USPTO directors called Kathi Vidal to change the proposed rules on terminal disclaimers, calling them “contrary to law” (Law360)

• The new IDS requirement includes a trap for the unwary: mistakes in filing the IDS could result in thousands of dollars and further delays (Foley)

• Chinese companies are rebranding with US subsidiaries and partners to get around restrictions around intellectual property (WSJ)

• Boies Schiller Flexner added three new patent litigators to its DC office (World IP Review)

• Micron got hit with $445 million in damages in a jury trial; Micron intends to appeal, in part arguing that one of the patents was ruled invalid by the PTAB (PC Gamer)

• A new ruling held that constitutional standing is not required for Section 337 cases (Dentons)

• WIPO adopted its first new treaty in over a decade regarding genetic resources and associated traditional knowledge, building on the GBD treaty and Nagoya protocol (PatentlyO)

• USPTO issued a memorandum instantly aligning its guidance to comply with the new obviousness test in LKQ v. GM, which overturned the Rosen-Durling test (USPTO)