Confusion in Trademarked NFTs

Abstract

The recent rise of the blockchain-based technology of non-fungible tokens (NFTs) has led to a number of lawsuits where NFTs are alleged to infringe trademarks. These cases have generally centered around the inquiry of likelihood of confusion, as most trademark cases do, but have also raised disputes as to the First Amendment and first sale doctrines and their relation to the likelihood of confusion inquiry. This article will discuss three such cases in detail, and draw some general insights about the likelihood of confusion inquiry and its interaction with these other doctrines, as well as make some more specific recommendations as to how these doctrines should be applied in the newer context of NFTs.

One of the primary recommendations of this article will be that courts should generally consider whether the alleged confusion would likely increase the amount that consumers would pay for the product at issue, particularly when evaluating survey evidence of actual confusion. Specifically with respect to NFTs, this may be likely because the value of NFTs is almost entirely based on “clout.” But when a true parody is at issue, this may be less likely, as consumers may not particularly care whether the parody was authorized by the brand that is being ridiculed. Alleged confusion that will not tend to affect consumer purchasing decisions should generally be discounted or at least given less weight.

Another more general recommendation will be that likelihood of confusion should be considered a question of law based on underlying facts, which is surprisingly contrary to the current approach of a majority of circuit courts that currently view the question as one of fact. This article also considers when the First Amendment based Rogers doctrine might apply to NFTs in light of the recent but limited guidance from the Supreme Court in Jack Daniel’s v. VIP Products, and makes recommendations as to how courts should approach the doctrine when it does apply. As for the first sale doctrine, this article argues that it should generally not protect NFTs linked to physical products, as the NFTs should generally be considered separate products and evaluated under the standard likelihood of confusion analysis.

Introduction

Non-fungible tokens, (“NFTs”) are a recent blockchain based technology that has many possible uses. One of the primary current uses of NFTs is for the sale of digital art. There are various ways to define an NFT, but basically, it is a unique blockchain-based token that may be associated with or linked to a specific piece of digital artwork or another type of file.

NFTs provide a way to create artificial scarcity in the digital art market.1 Roughly speaking, just as an artist might create only 100 signed authentic copies of a piece of physical artwork, an artist could create only 100 NFTs linked to a piece of digital artwork, meaning that there are only 100 “authentic” copies of that piece of digital artwork.2 One could say that NFTs in this context are akin to digital certificates of authenticity.3

NFT artists sometimes include aspects of others’ trademarks in their digital artwork, and a number of NFT creators have already been sued for trademark infringement.4 The central inquiry in trademark law is likelihood of consumer confusion,5 so the foundational question in these lawsuits is whether NFT buyers are likely to think that the NFT originated from or was otherwise endorsed by the trademark plaintiff.6

If sued for trademark infringement, the artist defendant might raise the defense that they are merely making a parody of the trademark, and their use of it should thus be protected by the First Amendment, invoking what is known as the Rogers doctrine.7 Indeed, this defense has already been raised in multiple trademark cases involving NFTs.8

The Supreme Court in Jack Daniel’s recently weighed in on the Rogers doctrine for the first time, and though it left much about the doctrine up for debate, the Court did cabin the doctrine, holding that it does not apply when the defendant is using, as its own trademark, aspects of the plaintiff’s trademark.9 But Rogers could potentially still protect some NFTs where the unauthorized use or parody of a mark has a less prominent role in the NFT.10

Given the newly limited scope of the Rogers doctrine, many NFT artwork cases will revert to the standard likelihood of confusion analysis. Yet that analysis itself is not well defined, with each circuit using a different multifactor test.11 There is not even a consensus amongst the circuits on whether the inquiry is ultimately one of fact or of law.12 The Supreme Court in Jack Daniel’s did recently offer some limited guidance on the likelihood of confusion inquiry, particularly in the context of parody, and also specifically with respect to the underlying factor of survey evidence of actual confusion, but again, much remains to be determined.13

Another defense that has been raised in cases involving NFTs is the first sale doctrine.14 The gist of this defense is that the NFT is being used merely as a technological tool to track the ownership of a physical item that has already been lawfully purchased by the trademark owner.15 The defense argues that the trademark rights with respect to both the physical item and the NFT were exhausted upon that first sale of the physical item. But the defense should not be available if the NFT is considered a separate product from the purchased item.16 This defense is also closely linked to likelihood of confusion and the question of whether purchasers would be confused into thinking that the NFT was created by the trademark owner.17

This article will explore how the likelihood of confusion analysis should apply in the context of NFTs, including interactions with the First Amendment and first sale defenses, in light of the Supreme Court’s recent guidance. This article will also use the relatively novel context of NFTs as a lens through which to view trademark law more generally, and will offer recommendations about some important open general questions in trademark law.

One of the primary normative recommendations will be that courts should pay more attention to whether confusion over the mark is likely to affect purchasing decisions, as in, would the buyer of the allegedly infringing product have paid less for it if they were not confused about whether the product was sponsored by the trademark owner.18 This may often be the case when NFTs are at issue, as the market for NFTs (like the market for artwork generally) is essentially a market for “clout,” and the unauthorized use of a prestigious trademark may add to the perceived clout and value of the NFTs. On the other hand, when a true parody is at issue, this may be less likely, as consumers would most likely not particularly care whether the parody was authorized by the brand that is being ridiculed.

This consideration should come into play for courts particularly in determining the weight to be given to survey evidence of actual confusion. Survey evidence of confusion that is unlikely to affect purchasing decisions should be given less weight in the ultimate balancing of factors.19 Relatedly, this article will argue that although the likelihood of confusion inquiry may involve underlying factual questions, the ultimate determination should be considered a legal question, which somewhat surprisingly is contrary to the current approach of a majority of circuit courts’ decision to treat the issue as a purely factual one. This recommendation is in line with some limited guidance that was offered by the Supreme Court in Jack Daniel’s.20

With respect to the First Amendment based Rogers doctrine, circuit courts are currently split on the correct approach, and the Supreme Court in Jack Daniel’s declined to resolve the circuit split.21 This article will recommend that when the doctrine does continue to apply, courts should use a flexible approach like that generally used in the Second Circuit, rather than a rigid approach such as that used in the Ninth Circuit.22

This article proceeds as follows. Part I provides an introduction to NFTs and how they interact with trademark law, discussing three recent trademark cases involving NFTs. Part II discusses the likelihood of confusion inquiry in general, and the First Amendment and first sale defenses, in light of the Supreme Court’s recent guidance. Part III makes some normative recommendations about how courts should apply the law of likelihood of confusion and these defenses to the emerging context of NFTs and in general. The article then briefly concludes.

1. Introduction to Trademarked NFTs

This part will provide a brief explanation of NFTs in general, and then discuss three cases where NFTs are allegedly infringing trademarks, introducing the legal issues that will be fleshed out further in subsequent parts, and their application in the NFT context.

A. What are NFTs?

Non-Fungible Tokens, or NFTs, are a somewhat recent blockchain based technology that raises some interesting and novel intellectual property issues. NFTs are stored on a blockchain, which operates as a public ledger, allowing ownership of the NFT to be transferred and authenticated electronically without the need for a physical item or a trusted third party, such as a bank.23 The blockchain is a distributed public ledger that keeps track of who owns what and is kept on computers across the world choosing to run the appropriate software.24 Each NFT is generally associated with a link to a digital content file, such as a picture or video.25

One of the primary uses of NFTs so far is to sell and track “ownership” of digital artwork. “Ownership” is in quotations because the owner of the NFT generally does not own the artwork in exactly the same way that the buyer of a physical painting owns that painting.

An NFT does not necessarily carry any intellectual property rights with it, but it is possible for NFTs to be set up to transfer intellectual property rights in the associated artwork.26 Although most NFTs do not include any IP rights, the NFT creator can choose to associate IP rights (to the extent the NFT creator has such rights in the first place) with the NFT, and some have done this.27 These IP rights could be specified in a content license that could be associated with the NFT.28 The content license could potentially be transferred to the new NFT owner via a “smart contract” that automatically executes every time the NFT is transferred.29

For example, an artist who owns the copyright in a piece of digital artwork could create NFTs that carry and automatically transfer to the NFT owner a license to display the artwork, or even a license to create derivative works. In a case like this, aside from just “clout,” the buyer of the NFT is receiving potentially valuable legal rights.

However, in a case where no intellectual property rights are carried with the NFT, which seems thus far to be the usual case, it is difficult to say what exactly the owner of an NFT owns. The market for NFTs has thus been called a market for “clout.”30 All the owner of the NFT necessarily owns is the NFT itself, which links to a picture of the digital artwork. Thus, the NFT owner generally has no right to stop others from copying the digital artwork, although others would not have the same claim to “authentic” ownership.31

The concept of “clout” is difficult to define precisely, but essentially it is the prestige that accompanies ownership of a scarce resource such as rare artwork. Bryan Frye explains that the art market “values the clout that accompanies ownership. Of course, art collectors value art for its own sake. Who doesn’t? But art isn’t scarce. Vast quantities are available for next to nothing. Only desirable art is scarce. Which is to say, only art with clout is scarce.”32

An NFT owner might, for example, publicly display their NFT on social media and, assuming they don’t own the copyright in the associate image, would generally lack the right to stop others from taking screenshots of the displayed artwork and displaying it themselves.33 Yet unlike those copiers, the NFT owner would have a digital NFT token signifying authenticity that the NFT owner could brag about on social media or potentially resell for a profit. Ownership of the “authentic” token could carry other benefits, as well; for example, some social media sites might choose to require proof of authentic ownership before allowing a user to display the NFT artwork on their profile on the site. The market for NFTs has fluctuated wildly and is largely driven by speculation. At the height (at least so far) of NFT mania in 2021 and early 2022, it was not that unusual for NFTs to sell for upwards of seven figures.34

All of this may be difficult to grasp in the abstract, but the following discussion of current lawsuits involving trademarked NFTs should be helpful. Trademark suits involving NFTs will likely continue to arise, as many established brands are now registering their marks specifically for NFTs.35 As discussed below, the fact that the value of NFTs are largely based on clout may be important for thinking about how trademark law should apply to NFTs.

In the cases discussed below, the allegedly infringed trademark itself carries some prestige as the better known brand as compared with that of the alleged infringer, and thus if the infringed mark were associated with the NFT, that would likely increase the clout and value of the allegedly infringing NFTs. If a purchaser were confused into thinking that the mark owner authorized or was responsible for the NFT, that may well significantly increase the amount that the purchaser would be willing to pay for the NFT, which this article will argue should tend to weigh in favor of likelihood of confusion and trademark infringement.

B. Yuga Labs v. Ryder Ripps

Yuga Labs created the popular Bored Ape Yacht Club (BAYC) set of NFTs. These NFTs gained popularity as celebrities such as Snoop Dogg publicly “Aped In” and bought them, helping to drive up the prices of the NFTs.36 Ryder Ripps is a “conceptual artist” who created his own NFTs linking to the exact same BAYC images that Yuga Labs used. Ryder Ripps claims this was a parody of sorts, aimed to educate the public about the nature of NFTs, and also about the allegedly racist and neo-Nazi aspects of the BAYC project.

Yuga Labs filed suit in the Central District of California, asserting trademark infringement against Ryder Ripps, based on his alleged unauthorized copying (or “counterfeiting”) of Yuga Labs’ BAYC NFTs.37 Each Yuga Labs BAYC NFT is linked to a digital AI generated artwork picture of a “bored ape.” The defendant, Ryder Ripps, allegedly copied the exact images (including the trademarks) and sold them under the very similar name RR/BAYC.38

One might wonder why Yuga Labs did not also assert a claim for copyright infringement, since Ripps’ wholesale copying of the original BAYC artwork would seem to present a fairly straightforward case for copyright infringement. The answer may be that the original BAYC images were generated by artificial intelligence and thus may not be eligible for copyright protection, or at least Yuga Labs did not want to implicate that issue.39

The following picture, taken from the Yuga Labs complaint, shows the original Yuga Labs BAYC NFT #1058 (each BAYC NFT is identified by a number) on the left, and the RR/BAYC “knockoff” NFT on the right (which uses the exact same identification number, #1058).

Some of the BAYC digital images contain trademarks for which Yuga Labs has applied for registration, and which have also been allegedly copied by Ryder Ripps.40 For example, the logos on the clothing (the hat on the left and the shirt on the right) worn by the bored apes in the following picture are allegedly trademarks of Yuga Labs that were copied by Ryder Ripps.

Since these trademarks on the bored ape images indicate the source of the NFTs (Yuga Labs), the fact that the associated artwork may have been AI generated would not seem to create a barrier for trademark protection. Thus embedding trademarks in AI generated artwork associated with NFTs may be (and in this case was) an effective strategy for obtaining some IP protection even if AI generated artwork is not subject to copyright protection.

Unlike many NFTs which carry no IP rights, the Yuga Labs BAYC NFTs are associated with a content license that grants the NFT holder fairly broad intellectual property rights not only to display the images but also to create derivative works using the Bored Ape Yacht Club trademarks, brand, and characters.41 Indeed, one restaurant entrepreneur reportedly bought a number of BAYC NFTs in order to obtain these rights and access to the BAYC creator support network, and then started a Bored Ape Yacht Club themed restaurant in Los Angeles.42

The original Yuga Labs BAYC NFTs, of which there are only 10,000 total, have become popular and have sold for up to millions of dollars each, prices that were driven largely by speculation but also perhaps to some extent by perceived value of the associated IP rights to use the bored ape characters and brand.43 Consumers trying to buy Yuga Labs BAYC NFTs might be confused into thinking that the Ryder Ripps “RR/BAYC” knock-off versions originated from Yuga Labs. The “RR / BAYC” NFTs created by Ryder Ripps carry no such IP rights because the knock-off NFTs do not come with a content license, and furthermore could not because these rights (such as the trademarks worn by the bored apes and any IP rights in the characters themselves) would be owned by Yuga Labs, not Ryder Ripps, and one cannot grant a license to something one does not own in the first place.

The legal standard for likelihood of confusion will be discussed in more detail in Part II, but to the extent that the allegations in the complaint are true, Yuga Labs seems to have a fairly straightforward and strong case for likelihood of confusion and trademark infringement, and as discussed below, Yuga Labs did in fact win at the district court on its trademark claim. The “goods” (NFTs) and the trademarks used are essentially identical,44 and the fact that the defendant added “RR” before BAYC is unlikely to save him.45

The situation, with BAYC being the more well-known and valuable NFT brand as compared with Ripps, suggests that the defendant Ripps may have been intentionally trying to cause consumer confusion, given the high prices that the original “authentic” BAYC NFTs were selling for, and there was evidence that some consumers were actually confused.46 In other words, the BAYC trademark carries a decent amount of clout given its popularity and celebrity endorsements, which would likely add to the perceived value or clout of the Ripps NFTs if people are confused into thinking that the Ripps NFTs are actually associated with the authentic Yuga Labs BAYC brand.

The BAYC marks would likely qualify as at least suggestive, giving them decent inherent strength.47 And given the novelty of NFTs as a concept, there is reason to think that a substantial number of consumers may be confused about what they are buying.48

Ripps asserted that his NFTs are a form of parody and should be protected against claims of trademark infringement under the Rogers doctrine, which is rooted in the First Amendment. 49 Ripps claims that Yuga has embedded racist “dog whistles” in some of the bored ape images and logos.50 Ripps also claims that the company’s name, Yuga Labs, is itself a reference to neo-Nazi and racist concepts.51 Ripps claims that he created the RR/BAYC project as a way to comment on and bring attention to the racism in the Yuga Labs project, and also to educate the public about the nature of NFTs.52

As evidence in support of this defense, Ripps claims that his customers must agree to an acknowledgement that these are new NFTs minted by him.53 The acknowledgement also explains that one cannot actually “copy” an NFT, as each NFT is a unique blockchain based token (though it could link to the same image as another NFT, as his do), and thus purportedly educates his buyers about the nature of NFTs.

Ripps’ First Amendment defense, though ultimately unsuccessful, raises some interesting questions. Initially, one might think that Ripps does not have much claim to First Amendment protection because he directly copied the BAYC images from the Yuga Labs version, and thus did not transform them or add any of his own creative content. But from a broader perspective, Ripps’ claim that his RR/BAYC project is a form of “appropriation art” intended to educate the public about the allegedly racist nature of the Yuga Labs project, and also about the nature of NFTs in general, could arguably have some merit and entitle him to some degree of First Amendment protection. For that to be the case, though, Ripps should have made these educational points more explicit. Ripps did not include much if any commentary in his NFTs explaining the racist nature of the Yuga Labs project or the nature of NFTs in general. If one has to read the court filings to understand the educational point of the project, perhaps Ripps should have featured these educational points more prominently, and his NFTs are then likely not transformative enough to warrant First Amendment protection.

Indeed, in an April 2023 order, the district court granted Summary Judgment for Yuga on likelihood of confusion, and the court (probably correctly) did not buy Ripps’ First Amendment Rogers defense.54 On likelihood of confusion, the court found that the goods (NFTs) and the trademarks were essentially identical, and even found that Ripps had engaged in a bad faith attempt to cause consumer confusion; that is, that “Defendants intentionally designed the RR/BAYC NFTs and sales website to resemble Yuga’s branding,” noting for example that the same labeling numbers (e.g., #1058 in the figure above) were chosen.55 On the First Amendment issue, the court held that Rogers did not apply because the “Defendants’ sale of RR/BAYC NFTs is no more artistic than the sale of a counterfeit handbag.”56

The Yuga Labs case thus gives a basic idea of how the central trademark issue of likelihood of consumer confusion can play out in the context of NFTs. The Hermes case discussed next provides another example with a stronger claim to First Amendment protection.

C. Hermes v. Rothschild

Mason Rothschild created 100 original digital images depicting fuzzy versions of the luxury “Birkin” handbags, selling this artwork via NFTs and calling them “MetaBirkins.”57 The NFTs “have sold for prices comparable to real-world Birkin handbags,” which sell “for anywhere from thousands of dollars to over one hundred thousand dollars.”58

When Hermes, the owner of the Birkin trademarks, sued Rothschild for trademark infringement in the Southern District of New York, Rothschild asserted a First Amendment defense stemming from the 1989 Second Circuit case of Rogers v. Grimaldi.59

As will be discussed in detail in Part II, the Rogers case reasoned that when allegedly infringing trademarks are used in an artistic context, courts should balance the “public interest in avoiding consumer confusion” against the “public interest in free expression” so as to avoid intruding on First Amendment values, and stated that this “balance will normally not support application of the [Lanham] Act unless the title has no artistic relevance to the underlying work, or . . . unless the title is explicitly misleading as to the source or the content of the work.”60

This “Rogers doctrine” has been adopted in about half of the circuits, and the two-part no artistic relevance or explicitly misleading test has been applied far outside of its original context.61 Artistic relevance is often said to be a “low bar” so most of the cases turn on whether the allegedly infringing work explicitly misleads consumers.62 The Second Circuit, however, has clarified that explicitly misleadingness is not actually required; rather, what is required is a “particularly compelling” case of likelihood of confusion to outweigh the First Amendment interests at stake.63

In denying Rothschild’s 12(b)(6) Motion to Dismiss based (largely) on the Rogers doctrine, the court found that Hermes had pled facts that could plausibly make out a sufficiently compelling case for likelihood of confusion.64 The court noted that consumers posting on the “MetaBirkins” Instagram page had expressed confusion as to whether Hermes was affiliated with the NFTs, and that several media outlets had mistakenly reported that Hermes was affiliated with the NFTs.65 The court also pointed to the strength of the Birkin mark and potential bad faith on the part of Rothschild, as other factors that could potentially support a compelling enough case for likelihood of confusion.66

Rothschild had argued that the MetaBirkins were parodying the Birkin bags by depicting them as covered in fur, thus criticizing “the animal cruelty inherent in Hermes’ manufacture of its ultra-expensive leather handbags.”67 A legitimate claim to parody would bring the MetaBirkins closer to the core of First Amendment protected expression and strengthen the Rogers defense.68 The court noted however that certain statements given by Rothschild in interviews suggested that the MetaBirkin NFTs were meant more as “a tribute” to the famous Birkin bags, rather than as a parody or criticism.69 The court thus concluded that given the disputed facts relevant to the confusion inquiry, the motion to dismiss should be denied.

Although not directly at issue in this case, the court did flag an issue that may arise soon, noting that if the MetaBirkins had been “virtually wearable” clothing (which they were not here), they might be treated more as ordinary or mundane commercial products than as artistic works, and thus might not be entitled to any First Amendment protection under the Rogers doctrine.70 However, courts have in the past applied Rogers in the context of virtual worlds such as video games, so the issue may turn on the degree to which artistic expression can be found in the virtual items or perhaps in the entirety of the virtual world itself.71

Rothschild continued to argue the Rogers First Amendment defense in a Motion for Summary Judgment.72 The motion referenced an expert declaration by Dr. Blake Gopnik, a “noted art historian” and critic, which purports to explain “how the images and NFTs produced and sold by Mason Rothschild find their natural and obvious home among the artistic experiments carried out by modern artists over the last century,” and how “Rothschild seeks to probe the nature of art, and of commerce, by blurring the distinction between the two categories.”73 In the motion Rothschild, referencing the report of Dr. Gopnik, explains that the MetaBirkin bag images are depicted in “goofy, garish fake fur,” that “flags the absurdist, parodic intent of the project and are both a fanciful tribute to the Birkin bag, which has become a cultural object signifying extreme wealth, and a reference to the fashion industry’s fur-free initiatives.”74 With respect to the Rogers test, Rothschild argued that Hermes cannot show explicit misleadingness and that there is not a “particularly compelling” case for confusion, given the differences in the marks (Birkin v. MetaBirkin) and the large differences in the products (actual bags versus NFTs depicting parodic or fanciful images of the bags).75

The court nevertheless rejected Rothschild’s Motion for Summary Judgment.76 In discussing which works are artistic enough to be considered under Rogers, the court stated that the “touchstone” of the inquiry “is whether the trademark was used to mislead the public about the origin of the product or the parties that endorse or are affiliated with it.”77 The court found again that Rogers did apply, because “Rothschild’s use of Hermes’ marks did not function primarily as a source identifier that would mislead consumers into thinking that Hermes originated or otherwise endorsed the MetaBirkins collection, but rather as part of an artistically expressive project.”78 However, the court thought that there were genuine issues of material fact regarding whether consumers were sufficiently likely to be confused to meet the higher (“particularly compelling”) threshold for likelihood of confusion even under Rogers.79

At trial, the jury ruled in favor of Hermes and found a sufficient likelihood of confusion. The jury was instructed to find the defendant liable, it must find that his “use of the Birkin mark was not just likely to confuse potential customers but was intentionally designed to mislead potential consumers into believing that Hermes was associated with Mr. Rothschild’s MetaBirkins project.”80 The court thus essentially interpreted the “explicitly misleading” prong of Rogers to mean “intentionally misleading.”81

The decision to give this question to the jury seems questionable, as it is ultimately a balancing of the likelihood of confusion against the First Amendment interest at stake, which would seem to be more appropriately treated as a question of law. Different interpretations of the Rogers doctrine amongst various courts and circuits will be discussed in detail in Part II.82

The Hermes decision, though it may be appealed, is consistent with a central argument of this article. Since Hermes is a luxury brand which carries a good deal of clout, it seems that consumers who think that the MetaBirkin NFTs are associated with Hermes would likely pay more for those NFTs. In other words, the consumer confusion at issue would seem likely to affect purchasing decisions by adding to the perceived clout and value of the NFTs.

D. Nike v. StockX

Another area in which NFTs are raising some novel issues and also providing an interesting lens through which to examine old doctrine is the interaction between trademark law and the first sale doctrine. This is illustrated by the Nike v. StockX case before the federal district court for the Southern District of New York.

The defendant, StockX, operates an online resale platform and began selling NFTs depicting pictures of various Nike sneakers.83 Nike brought suit for trademark infringement, but StockX raised a first sale doctrine defense claiming that each NFT is “backed” by an actual pair of lawfully purchased Nikes that are stored in a “vault,” and that the owner of the NFT can at any time exchange it for the actual pair of shoes.84

StockX claims that generally upon each sale of an item on its re-sale platform “the seller ships the product to one of StockX’s eleven authentication centers around the world” where it is authenticated and then shipped to the buyer.85 StockX claims however that it has “attracted a significant number of customers who are interested in acquiring and trading current culture products, without any interest in immediately or ever wearing (or ‘consuming’) those products or taking physical possession of those products.”86 As a response to such customers, StockX claims that it began using NFTs as a technological tool to facilitate such trading of culture products, without requiring that the owner actually take possession of the physical product.

The buyer of a Nike shoe bearing Nike trademarks generally has the right to wear and resell the shoe under the first sale doctrine but, of course, does not have the right to manufacture new products bearing the Nike marks. This is likely part of the reason why StockX attempts to emphasize its NFTs “are absolutely not ‘virtual products’ or digital sneakers,” but rather are simply “claim ticket[s]” for the underlying items stored in the vault.87

StockX argues that because the NFTs are not separate products from the underlying shoes, its use of the NFTs as a technological tool to facilitate trading of the lawfully purchased shoes is protected by the first sale doctrine. One problem for StockX though is that the NFTs have sold for amounts far higher than the value of the actual Nike shoes, which tends to suggest that consumers do consider them separate products. StockX argues that this is not its fault, as it sets the initial price of the NFT as the same price as the underlying item, and cannot control the price of the NFTs after it releases them.88

But still, the fact that consumers seem to value the NFTs differently from the underlying shoes suggests that consumers view the NFTs as a separate product from the shoes. On the other hand, one might try to argue that if the NFTs ultimately sell for higher prices than the underlying item, this is due to the technological value add of the NFT which makes it easier to keep track of and resell the underlying items, but this is not inconsistent with the NFTs being merely a technological tool to track ownership.

At the end of the day though, the first sale argument seems like a stretch, so the case will thus likely turn on whether consumers are likely to be confused into thinking that Nike makes or sponsors the NFTs. The answer may depend partially on how StockX markets the NFTs. To the extent StockX is clear (as it claims to be) to consumers about the platform being a resale platform and the NFTs merely being used for authentication or to redeem physical goods, and makes it clear to consumers that the NFTs are not affiliated with Nike, this may help it avoid trademark liability by showing that there is no likelihood of confusion.89

Applying a primary argument of this article, courts should be aware of whether consumers who are confused into thinking that the NFTs were sponsored by Nike would likely pay more for those NFTs. In other words, if there is confusion, is it relevant to consumer purchasing decisions? The answer here may well be yes, given that the Nike brand itself carries a good deal of clout, at least as compared with the StockX brand, and thus would likely increase the clout and perceived value of the NFTs. This should be considered in the likelihood of confusion analysis. As far as where and how this should be incorporated into the analysis, that will be discussed in further detail in subsequent parts.

2. Confusion, First Amendment, and First Sale in Trademark Law

This part will go into more detail about the general law of trademarks in the areas raised by the NFT cases discussed above—specifically, likelihood of confusion and its relation to the First Amendment and the first sale doctrine. The legal issues raised by the NFT cases will be fleshed out and discussed in more detail, setting the stage for the normative recommendations that will be provided in Part III.

A. Likelihood of Confusion in Trademark Law

The central question in trademark law is likelihood of confusion; that is, whether the allegedly infringing items are likely to confuse a substantial number of consumers into incorrectly thinking that the items in question originated from or were endorsed by the trademark holder.90 This inquiry upholds the basic purposes of trademark law, which are to protect the brand and established goodwill of trademark holders against free riders, and to protect the public against confusion as to the source of goods and services.91 Accordingly, the Supreme Court has recently referred to consumer confusion as the “bete noire of trademark law,” because it “stands directly opposed to the law’s twin goals of facilitating consumers’ choice and protecting producers’ good will.”92

There is a fair amount of confusion surrounding likelihood of confusion, even though it lies at the very heart of trademark law.93 The test for likelihood of confusion unfortunately varies by circuit, but there are 5 or 6 “core” factors that can be found in most of the tests:

1. The strength of the plaintiff’s trademark;

2. The degree of similarity between the plaintiff’s and defendant’s goods or services;

3. The degree of similarity in the marks used by the plaintiff and defendant;

4. Whether the defendant’s intent is to cause confusion (bad faith);

5. Evidence of actual confusion;

6. The level of consumer sophistication in the marketplace.94

Almost every factor has been referred to, at one time or another, as the “most important” factor.95

To make matters worse, there is not even a consensus amongst the circuits on whether the likelihood of confusion inquiry is ultimately one of fact or of law.96 Circuit courts are currently split on this important question. The Second Circuit takes what seems like the correct approach (as will be argued in Part III(B), infra), treating some individual factors as factual but others as legal and the ultimate question of balancing them as legal.97 The Federal Circuit and the Sixth Circuit also follow this approach.98 However, the majority of circuits—including the Ninth Circuit—consider the ultimate question of likelihood of confusion to be one of fact.99 Even within circuits, courts can be inconsistent on which approach they take to this fundamental question.100

Without resolving the circuit split, the Supreme Court recently provided some limited guidance on this question and on some other aspects of likelihood of confusion in the Jack Daniel’s case.101 The case involved the “Bad Spaniels Silly Squeaker” dog toy, which the district court found confusingly resembled a bottle of Jack Daniel’s Old No. 7 Black Label Tennessee Whiskey, ruling in favor of Jack Daniel’s on the claim for trademark infringement.102

The dog toy contains a number of prominent differences from the Jack Daniel’s bottle, including a picture of spaniel, and jokes such as “Old No.2 on your Tennessee Carpet,” instead of “Old No. 7 Whiskey.” The dog toy had other strong indications that it was a parody; for example, replacing the alcohol content descriptions with “43% POO BY VOL.” and “100% SMELLY.”103 The district court’s finding of likelihood of confusion thus seems questionable.

The products are quite different—dog toys versus whiskey—so this factor should have weighed strongly against confusion.104 Yet the district court weighed this factor in favor of Jack Daniel’s, because Jack Daniel’s had licensed its marks for use on a wide variety of products, including dog leashes, dog collars, and dog houses.105 This seems highly questionable, the public associates Jack Daniel’s with whiskey, not dog products, and other courts have appropriately held that limited licensing of this sort does not suffice to tilt the proximity of the products factor in favor of confusion when the core product lines are far apart.106

The marks are somewhat similar but not identical; VIP used the name “Bad Spaniels” rather than “Jack Daniels” and embedded various jokes on the dog toy, making it fairly clear that the dog toys were parodies.107 The defendant’s intent was not to cause confusion, but rather to create a humorous parody. On this point, the district court seems to have committed legal error by finding that the factor favored Jack Daniels.108 A successful parody will generally not cause confusion, and VIP’s intent was to create a parody, not to create confusion.109 Indeed various courts have appropriately recognized that “[a]n intent to parody is not an intent to confuse the public.”110

Although Jack Daniel’s is a strong mark, the weight given to this factor should generally be diminished in a case of parody, for courts have recognized that it “is a matter of common sense that the strength of a famous mark allows consumers immediately to perceive the target of the parody, while simultaneously allowing them to recognize the changes to the mark that make the parody funny or biting.”111

The district court, in finding likelihood of confusion, relied heavily on survey results that found that 29% of purchasers were likely confused.112 Although this might be enough to favor likelihood of confusion in some cases, it should not be enough to establish a sufficiently compelling case for confusion in a case where many of the other factors should weigh against confusion. Indeed, in the Rogers case, discussed infra, there was survey evidence indicating a higher rate of confusion, but the court nevertheless found no trademark infringement.113

Moreover, it is questionable whether the confusion that was measured in the surveys is the type that would tend to affect consumer purchasing decisions. Buyers of dog toys are unlikely to think they are buying whiskey, and probably would not give much thought to or care whether the obviously parodic dog toy they are buying was officially endorsed by the Jack Daniel’s company or not. As such, even if consumers are confused into thinking that Jack Daniel’s endorsed the dog toy, this confusion should be given less weight to the extent that it is unlikely to have affected purchasing decisions.114 In light of the major differences between the products, the differences in the marks, and the defendant’s clear parodic intent, a court could potentially find that the district court erred as a matter of law in finding that consumers are likely to be confused as to the source of the dog toy.115

Or at least, a court could easily find that in circuits where the ultimate question is considered one of law, as this article argues it should be. The Ninth Circuit, however, treats the confusion issue as one of fact, so although the appellate court may have been rightly skeptical about the district court’s likelihood of confusion finding, the court reversed on other grounds. On appeal, Bad Spaniels argued that the dog toy was an expressive parody entitled to First Amendment protection under Rogers, and the Ninth Circuit agreed that the district court erred in finding trademark infringement without first requiring the plaintiff to satisfy at least one of the two Rogers prongs (no artistic relevance or explicitly misleading).116 The Ninth Circuit did not otherwise directly address the district court’s likelihood of confusion analysis.

The Supreme Court reversed the Ninth Circuit, finding that Rogers did not properly apply to the case for reasons that will be discussed further in the next subsection.117 The Court found that the case thus should turn on likelihood of confusion and remanded for a re-evaluation of likelihood of confusion, providing the guidance that the Bad Spaniels toy’s expressive or parodic message “may properly figure in assessing likelihood of confusion.”118 The Court explained that the likelihood of confusion inquiry “is not blind to the expressive aspect of the Bad Spaniels toy” because “consumers are not so likely to think that the maker of a mocked product is itself doing the mocking.”119 In other words, “to succeed, the parody must also create contrasts,” and “once that is done (if that is done), a parody is not often likely to create confusion.”120

Unfortunately, the Supreme Court did not answer the question of whether likelihood of confusion is a question of law or fact, but it did at least note that the issue of likelihood of confusion need not always go to trial and is potentially susceptible to disposition at the summary judgment or motion to dismiss stage.121 This statement seems to at least provide some support for treating the ultimate inquiry as one of law, as the Second Circuit does.

The factor of evidence of actual confusion has become somewhat problematic, as courts have sometimes given the factor an outsized role, relying on surveys which purport to show that a certain percentage of consumers are confused. This may be due to the amorphous nature of the multifactor likelihood of confusion balancing tests, causing courts to latch on to the seemingly more concrete and quantitative survey evidence; i.e., if the percentage of consumers confused in the surveys is high enough, courts will tend to find a likelihood of confusion. Indeed in the Jack Daniel’s case, the district court’s finding of likelihood of confusion relied heavily on survey evidence of confusion, even though the surveys were questionable and other factors should have weighed against confusion.122 Courts and commentators alike have discussed the many problems with survey evidence of confusion, and how they tend to overestimate the likelihood of consumer confusion, in part because “people are more careful when they are laying out their money than when they are answering questions.”123

VIP argued before the Supreme Court that all the survey evidence really showed was that consumers were confused about what the law required. That is, consumers may have thought that VIP got Jack Daniels’ permission for the dog toy only because they mistakenly thought that such permission is always required for parodic uses of trademarks. Some of the answers to the surveys supported this contention.124 More generally, these surveys are expensive, and a party with resources can hire experts that are able to craft the questions and present the results to support the desired outcome.

Although the Supreme Court in Jack Daniel’s did not wade too deeply into the likelihood of confusion analysis, a concurrence by Justice Sotomayor, joined by Justice Alito, cautioned courts should not to rely too heavily on survey evidence of confusion (as the district court in the case likely did) and to be cognizant of the potential problems with such surveys, recognizing that they are but one factor in the analysis. The concurrence made the point that “in the context of parodies and potentially other uses implicating First Amendment concerns, courts should treat the results of surveys with particular caution.”125 Justice Sotomayor emphasized that survey evidence “should be understood as merely one piece of the multifaceted likelihood of confusion analysis,” and that survey answers “may reflect a mistaken belief among some survey respondents that all parodies require permission from the owner of the parodied mark.”126 She cautioned that courts “should be attentive to ways in which surveys may artificially prompt such confusion about the law or fail to sufficiently control for it.”127

Normative recommendations for how the likelihood of confusion test should apply in the context of NFTs, with a particular focus on issues related to survey evidence of actual confusion, will be discussed further in Part III.

B. The First Amendment and the Rogers Doctrine

This subsection will discuss the origin of the Rogers doctrine granting First Amendment protection to expressive uses of trademarks, as well as how the doctrine has been developed or rejected in various circuits, and recent Supreme Court guidance on the doctrine. This doctrine was raised as a defense in two of the three NFT trademark cases discussed in Part I. Although the Supreme Court did recently cabin the doctrine in the Jack Daniel’s case, the doctrine will still at least arguably apply in some NFT cases and so likely remains important for NFTs.

To briefly review the facts of Rogers, Ginger Rogers and Fred Astaire had established themselves as “among the most famous duos in show business history,” through their “incomparable performances in Hollywood musicals.”128 In March 1986, the defendants produced a film entitled “Ginger and Fred,” directed by Federico Fellini, which “tells the story of two fictional Italian cabaret performers, Pippo and Amelia, who, in their heyday, imitated Rogers and Astaire and became known in Italy as ‘Ginger and Fred.’”129

Shortly after distribution of the film began, Ginger Rogers brought suit seeking injunctive relief and money damages, alleging that the defendants violated section 43(a) of the Lanham Act by creating the false impression that the film was about her or that she sponsored or was somehow involved in it, and also violated her common law right of publicity.130 The case thus presented a “conflict between Rogers’ right to protect her celebrated name and the right of others to express themselves freely in their own artistic work.”131 Attempting to avoid intruding on First Amendment values, the court stated: “in general the [Lanham] Act should be construed to apply to artistic works only where the public interest in avoiding consumer confusion outweighs the public interest in free expression.”132

Having first set forth this generally applicable balancing framework, the court then tentatively ventured a more specific test for the very specific situation at issue, stating that in “the context of allegedly misleading titles using a celebrity’s name, that balance will normally not support application of the Act unless the title has no artistic relevance to the underlying work whatsoever, or, if it has some artistic relevance, unless the title is explicitly misleading as to the source or the content of the work.”133

The Rogers court thus provided a basic balancing framework for artistic works generally (avoiding consumer confusion v. free expression), and then attempted to sketch a more specific two-part test for how that balancing framework would “normally” apply in the context of titles of artistic works using celebrity names. This two-part initially conjectural test of no trademark liability unless (1) no artistic relevance, or (2) explicitly misleading, has taken on a life of its own, and has sometimes been rigidly applied far outside of the context that the test was originally explicitly limited to.

Judge Griesa concurred with the result reached by the majority opinion (finding the likelihood of confusion outweighed by the relevant free expression interests), but had “substantial disagreement with the opinion otherwise.”134 Judge Griesa thought that the majority’s distinction between explicit and implicit misleadingness would “prove to be unsound and unworkable,” and that it “should be left to future courts, dealing with real cases, to determine if there are to be exceptions to the First Amendment protection which would seem to be generally afforded to artistically relevant titles.”135

Although the general Rogers balancing framework of free expression against likelihood of confusion has appropriately been widely applied, the Second Circuit has appropriately not generally applied the more specific no artistic relevance or explicitly misleading test. Instead, the court has essentially walked the explicitly misleading prong back towards a likelihood of confusion analysis, albeit one where a stronger or more compelling likelihood of confusion is required to outweigh the free expression interests at stake. This balancing approach presumably contemplates a sliding scale: the stronger the free expression interests at issue are, the stronger the case for likelihood of confusion must be.

The Second Circuit first extended the Rogers balancing framework beyond the context of titles to apply as well to the content of artistic works in Cliffs Notes, stating: “we hold that the Rogers balancing approach is generally applicable to Lanham Act claims against works of artistic expression, a category that includes parody.”136 But importantly, in referring to the “Rogers balancing approach,” the court appears to have been referring to the general approach of balancing the public interest in avoiding consumer confusion against the public interest in free expression, rather than the more specific no artistic relevance or explicitly misleading language that Rogers proposed specifically for allegedly infringing titles using celebrity names.

This distinction is the source of some confusion because courts refer to a general and specific concept as the Rogers doctrine without specificity. The Cliffs Notes court stated that the more general balancing approach takes into account likelihood of confusion but “allows greater latitude for works such as parodies, in which expression, and not commercial exploitation of another’s trademark, is the primary intent.”137 In other words, “somewhat more risk of confusion is to be tolerated when a trademark holder seeks to enjoin artistic expression such as a parody.”138 The court thus essentially applied the standard likelihood of confusion analysis with an additional thumb on the scale for the defendant given the artistic character of the allegedly infringing work.139 In finding an insufficient likelihood of confusion as a matter of law, the court considered for example the degree of care which purchasers would normally exercise, a factor relevant to likelihood of confusion but irrelevant to explicit misleadingness.140

A few years later in Twin Peaks, this more flexible approach was endorsed by the very judge that wrote the Rogers opinion, Judge Jon Newman of the Second Circuit.141 Interestingly, the Twin Peaks court very briefly stated the more specific Rogers test, but dropped the word “explicitly,” finding artistic relevance and then stating: “the question then is whether the title is misleading in the sense that it induces members of the public to believe the Book was prepared or otherwise authorized by TPP.”142 The court then explained that this determination “must be made, in the first instance, by application of the venerable Polaroid factors,” (the Second Circuit’s factors for likelihood of confusion initially articulated by Judge Friendly), with the modification that “the finding of likelihood of confusion must be particularly compelling to outweigh the First Amendment interest recognized in Rogers.”143

This formulation of the Rogers balancing framework has subsequently been generally applied in the Second Circuit, although courts sometimes use the words “sufficiently compelling” instead of “particularly compelling.”144 Or, as explained in the Hermes case, “the most important difference between the Rogers consumer confusion inquiry and the classic consumer confusion test is that consumer confusion under Rogers must be clear and unambiguous to override the weighty First Amendment interests at stake.”145

Although the Second Circuit has backed away from the “explicitly misleading” formulation of Rogers, the Ninth Circuit still purportedly applies it. But when analyzing whether an expressive use of a mark is “explicitly misleading,” even courts in the Ninth Circuit often seem to focus on factors that relate more to whether the defendant’s use of the mark is likely to cause confusion. The E.S.S. case provides an example. 146 The case involved the video game “Grand Theft Auto,” set in the fictional cities of Los Santos, San Fierro, and Las Venturas, which are stylized versions of Los Angeles, San Francisco, and Las Vegas, respectively.147 The trademark suit centered around a virtual strip club in the game called “Pig Pen” in “East Los Santos,” which was loosely based on the actual strip club in East Los Angeles called “Play Pen Gentlemen’s Club.”148 The Ninth Circuit affirmed the district court’s grant of summary judgment in favor of the defendants under the Rogers doctrine, because it found the strip club artistically relevant to the video game, and not explicitly misleading.149 On the explicitly misleading prong, the court found it unlikely that “the Game would confuse its players into thinking that the Play Pen is somehow behind the Pig Pen or that it sponsors Rockstar’s [Grand Theft Auto] product.”150

The court’s observation that consumers are highly unlikely to be confused into thinking that “a company that owns one strip club in East Los Angeles . . . also produces a technologically sophisticated video game,”151 seems correct, but it also suggests that the truly relevant question is whether there is a sufficient likelihood of confusion. That is, the relevant question is not whether any misleadingness is explicit versus implicit; rather, the relevant question relates to the extent to which consumers are likely to be misled.

Other cases in the Ninth Circuit similarly look at factors sounding in confusion when purportedly applying the “explicitly misleading” prong of Rogers.152 These cases thus seem to validate Judge Griesa’s concern in his Rogers concurrence that the distinction between explicitly and implicitly misleading would prove unworkable,153 as even the Ninth Circuit does not really focus on the distinction, despite purporting to do so.

The Fifth Circuit has also adopted Rogers, and appears to have followed the lead of the Second Circuit in moving away from the “explicitly misleading” inquiry, in favor of requiring a particularly compelling likelihood of confusion to outweigh the public interest in free expression where use of the mark is relevant to an artistic or parodic work.154 The other two circuits to have adopted the Rogers doctrine, the Sixth and Eleventh Circuits, appear to apply the explicitly misleading requirement more rigidly, even outside the context of titles, although the doctrine in both circuits is relatively undeveloped.155

The Third, Fourth, Seventh, Eighth, and Tenth Circuits have not adopted the more specific Rogers doctrine (requiring no artistic relevance or explicit misleadingness) despite having had opportunities to do so, and some of these circuits have offered criticism of at least the strongest versions the doctrine, suggesting that it may be unnecessary because the likelihood of confusion test itself can sufficiently account for First Amendment interests.156

The Fourth Circuit has suggested that a proper application of the Lanham Act should generally be enough to protect free speech,157 explaining that when marks are “used to parody, satirize, criticize, comment or compare,” that use should be considered in and infuse the application of the likelihood of confusion factors, generally weighing against confusion.158 For example, in Louis Vuitton, the Fourth Circuit found that dog toys loosely resembling small Louis Vuitton handbags were not likely to confuse consumers into thinking that they were made or sponsored by Louis Vuitton, in part because the dog toys were “successful parodies of LVM handbags and the LVM marks and trade dress.”159 The court explained that a “parody relies upon a difference from the original mark, presumably a humorous difference, in order to produce its desired effect,” such that although a finding of parody is not dispositive as to likelihood of confusion, “an effective parody will actually diminish the likelihood of confusion, while an ineffective parody does not.”160

The Tenth Circuit has also suggested that the more specific Rogers test is unnecessary because trademark law already has a "built-in mechanism” to avoid First Amendment concerns given that “the Lanham Act requires proof of a likelihood of confusion. In the case of a good trademark parody, there is little likelihood of confusion, since the humor lies in the difference between the original and the parody.”161 A district court in the Tenth Circuit has criticized the test as “needlessly rigid and failing to account for the realities of each situation.”162

The Supreme Court in Jack Daniel’s had a chance to resolve the circuit split about the correct test for the Rogers doctrine, if the doctrine is valid at all, but declined to resolve this controversy. Instead, the Court ruled very narrowly and held that Rogers does not apply in the situation at hand, because the plaintiff’s mark is being used “as a mark” by the defendant.163 In similar words, the Court stated that when the defendant’s use “is ‘at least in part’ for ‘source identification’ . . . Rogers has no proper role.”164

As such, the Court vacated the Ninth Circuit’s decision applying Rogers. The Court made clear that it was taking no position on the Rogers doctrine in general, but noted that the doctrine only applies to “cases involving ‘non-trademark uses’—put another way, cases in which ‘the defendant has used the mark’ at issue in a ‘non-source-identifying way.’”165 The Court did appear to agree with the Fourth and Tenth Circuits that the likelihood of confusion analysis itself can and should take into account any parodic nature of the allegedly infringing work, and remanded for a reconsideration of likelihood of confusion in the case at hand with more attention given to such parodic aspects, but the Court did not resolve the question of whether Rogers might provide some additional protection in cases where it is applicable.

Whether, when, and how Rogers should continue to apply to NFTs and in general in light of the Jack Daniel’s decision will be discussed further in Part III.

C. First Sale Doctrine in Trademark Law

The first sale (or exhaustion) doctrine exists in patent, copyright, and trademark law, and generally provides that the lawful purchaser of an item has the right to use that item for its ordinary, intended uses, and to resell the item, without infringing intellectual property rights.166 One way of rationalizing the doctrine is that the intellectual property owner has “received their reward” upon the first sale and thus no further compensation is needed.167 The first sale doctrine is also said to facilitate resale markets and the smooth flow of goods through commerce, reflecting the common law’s antipathy to restraints on the alienation of chattels.168 It is sometimes thought of as the boundary between intellectual property law and commercial law, as the intellectual property rights “exhaust” upon the first sale.169

The doctrine does, of course, have its limits. The buyer of a trademarked item generally does not receive the right to create other products bearing the trademark, or to materially alter the trademarked product.170 Similarly, the buyer of a patented computer or widget has the right to use, display, and resell these items, but generally does not have the right to copy or reproduce them, for example.171 And in copyright law, the first sale gives the buyer the right to resell the copyrighted item and to display it locally, but not necessarily the right to make additional copies of the item.172

In general, courts have applied the first sale doctrine in trademark law to hold that “the right of a producer to control distribution of its trademarked product does not extend beyond the first sale of the product,” or in other words, “resale by the first purchaser of the original article under the producer’s trademark is neither trademark infringement nor unfair competition.”173 Courts have applied at least two exceptions to the first sale doctrine. The “Quality Control” exception applies when goods do not conform to the trademark holder’s quality control standards.174 And the “Material Differences” exception applies if the goods differ in a way that would likely be relevant to a consumer’s decision to purchase them.175

Both of these exceptions are closely tied to likelihood of confusion, as is the first sale doctrine in general. When a product has been materially altered or no longer conforms to the trademark holder’s standards, it could cause confusion, harming the trademark owner’s brand.176 Accordingly, courts have said that a difference is material if “consumers [would] consider [it] relevant to a decision about whether to purchase a product.”177 But when a product is simply resold in its normal form, no such confusion is likely. Courts have explained that reselling under the first sale doctrine is permitted because “confusion ordinarily does not exist when a genuine article bearing a true mark is sold.”178

For example, in Ledo Pizza, in addressing the first sale doctrine, the court explained that the “relevant question here” is whether the reheating of Ledo pizzas by Expressions Catering was likely to cause confusion as to the quality consumers would expect from pizzas purchased at the Ledo Pizza restaurants. The court ultimately found that there was no evidence to show that guests being served by Expressions Catering could have known that the pizzas being served were simply reheated Ledo pizzas, and thus there was no likelihood of confusion, and Expression Catering’s lawful purchase of the pizzas shielded them under the first sale doctrine.179

Thus, like the Rogers doctrine and its First Amendment protections, the first sale doctrine in trademark law is also closely tied to the likelihood of confusion analysis, to the extent it is not entirely clear that a separate doctrine is necessary.180 If the buyer of a Nike shoe were to add wheels to the shoes and then resell them, this would likely not be protected under the first sale doctrine because it could confuse consumers into thinking that the altered shoes were made by Nike when in fact they were not, and would fall into the material alteration exception to the first sale doctrine.181

Like the Rogers doctrine though, the first sale doctrine arguably provides added protection that is not as dependent on the loosely defined and unpredictable likelihood of confusion inquiry, and thus can potentially reduce litigation costs by allowing for earlier dismissal for the defendant.182 Someone who simply re-sells their lawfully purchased Nike shoes at a yard sale, if they were sued by Nike, could likely have the case quickly dismissed under the first sale doctrine without having to engage in potentially burdensome and sometimes uncertain litigation on the likelihood of confusion factors.

Recommendations for how these general first sale doctrine principles should be applied in the context of NFTs will be discussed further in Part III.

3. Normative Recommendations

This part will make some recommendations about how the doctrines discussed in this article should apply to NFTs and in general, in light of recent Supreme Court guidance.

A. Applying Likelihood of Confusion to NFTs

A contrast between the Hermes case and the Jack Daniel’s case helps to highlight a primary argument of this article, that courts should be cognizant of whether confusion as to source would likely increase the amount that consumers would pay for the product, and that this increased payment may be particularly likely when NFTs are at issue.

In the Jack Daniel’s case, it seems unlikely that many people were buying the Bad Spaniel’s dog toy or paying more for it because they were confused into thinking that it is sponsored by Jack Daniel’s. In other words, even if there is confusion about whether Jack Daniel’s endorsed the dog toy, which there may well be, it seems unlikely that such confusion would affect consumer purchasing decisions, so it should be considered irrelevant, or at least given less weight in the ultimate balancing analysis.183 It is hard to imagine many purchasers of the dog toy being upset to find out that it was not in fact sponsored by Jack Daniel’s, even if they had been confused into thinking that it was.184 Prestige or clout is not particularly important when dog toys are at issue, and moreover, although Jack Daniel’s is a well-known whiskey brand, it is not exactly a luxury brand with the prestige or clout of the Hermes brand.

By contrast, in the Hermes case, it seems quite plausible that some consumers were paying more for the NFTs because they thought that they were buying the “official” Hermes MetaBirkin NFTs. Hermes is of course a luxury brand with a good deal of prestige, and that matters a lot for NFTs, whose value is largely if not entirely based on clout. Similarly, it is quite plausible that confused consumers would pay more for RR/BAYC NFTs because they think they are buying the “authentic” Bored Ape Yacht Club NFTs. NFTs are essentially collector’s items being bought for their potential resale value or bragging rights, and both the Hermes brand and the BAYC brand have a good deal of the clout from which NFTs derive their value.185

Regarding the analysis, this clout should be considered primarily when evaluating the evidence of actual confusion. This is consonant with the warnings in Justice Sotomayor’s Jack Daniel’s concurrence for courts to closely scrutinize survey evidence of actual confusion and to be cognizant of whether it merely reflects confusion about the law.186 It was likely improper for the district court in the Jack Daniel’s case to rely heavily on survey evidence of confusion in finding a likelihood of confusion, as that survey evidence likely just showed that consumers were confused about what the law requires.187 That is, some consumers thought that the law generally required that the maker of a parodic dog toy needs to get permission from the brand that is being parodied, but the mere fact that they were confused about that should not make it so. If the confusion shown by survey evidence is not the type of confusion that would likely affect consumer purchasing decisions, the weight given to this factor in the ultimate balancing should be greatly reduced, as it is just one factor in the larger analysis.188

The degree to which confusion would affect purchasing decisions could also potentially be considered in the factor of the defendant’s intent, as this would provide an incentive for the defendant to intentionally cause confusion. It seems much more plausible that Ripps and Rothschild were intentionally trying to make consumers think that their NFTs were sponsored by BAYC or Hermes, respectively, as compared with VIP intentionally trying to make consumers think that Jack Daniel’s endorsed the dog toy, because such confusion would likely increase the value of the NFTs but not the dog toy.

Regardless, one need fret too much over which factor this consideration falls in with. Although the various factors can be a helpful tool, courts should keep in mind that the ultimate question is whether consumers are likely to be confused, and the most relevant confusion is that which would likely affect consumer purchasing decisions.

B. Confusion As a Matter Of Law

Recognizing likelihood of confusion as a question of law based on underlying facts would align the doctrine with similarly central and often dispositive inquiries in patent law and copyright—namely, obviousness and fair use, which the Supreme Court has made clear are both questions of law based on underlying facts.189 In clarifying that the ultimate question of fair use is one of law, the Court recently explained that the question “primarily involves legal work,” although certainly it may “involve determination of subsidiary factual questions.”190 Similarly, although likelihood of confusion involves some underlying factual inquiries (such as evidence of actual confusion) some factors involve primarily legal work, as does the ultimate balancing of the factors. For example, evaluating the strength or distinctiveness of the mark on the Abercrombie scale has been correctly said to involve primarily legal work.191 Like fair use, the Abercrombie scale for trademark distinctiveness was “originally a concept fashioned by judges,”192 as were the multi-factor balancing tests that each circuit has developed for evaluating likelihood of confusion.

The question of whether consumers are likely to be confused is no more inherently factual than whether an invention would have been obvious at the time it was made, or whether a use of another work is fair, yet for some reason likelihood of confusion is considered a factual question in the majority of circuits. Although the question of whether confusion is likely might (for some reason) initially perhaps seem more factual, commentators have recognized that the distinction between law and fact is ultimately a myth, and must be decided based on functional considerations.193 The question courts are ultimately asking is not a binary “is confusion likely or not.” The ultimate question is rather: is confusion likely enough to justify a claim for trademark infringement. This question involves matters of policy and balancing a variety of factors, some of them factual and some of them legal, and functionally is better suited for courts, who can try to conduct the balancing in accord with precedent and develop a greater degree of consistency and predictability in how the balancing is conducted.

Moreover, as others have observed, the question of likelihood of confusion may present a degree of circularity.194 This is true as well in the context of NFTs. If the law moves in the direction of saying that unauthorized use of trademarks in NFTs is infringing, it may come to be understood by consumers that NFTs bearing marks generally originate from or are endorsed by the mark owners. The Hermes and Yuga Labs cases discussed in Part I may already be pushing us in this direction.195 On the other hand, if the law moves in the opposite direction, it may eventually come to be understood that NFTs bearing trademarks do not necessarily have anything to do with the mark owner. At this point, with NFTs being fairly new, consumers may not yet know what to expect. The question of whether consumers in the public are likely to be confused is to some extent a cultural question, and thus provides an example of culture and law influencing each-other simultaneously.196

Thus, when new situations arise, such as with NFTs, where the law has not yet chosen a firm path, the likelihood of confusion question is to some degree a normative question involving legal policy judgments. As a practical matter, these policy judgments should be considered as a matter of law, albeit one that does also involve underlying factual considerations. For example, although evidence of actual confusion is factual evidence, the relative weight to be given such evidence in the ultimate balancing should be a legal question. Judges are experts in law, and are better suited than lay jurors to be making normative decisions about the direction the law should take.

It is sometimes argued that easy dismissal under Rogers is necessary because litigating likelihood of confusion to trial would prove prohibitively costly for many artistic defendants.197 But this concern could potentially be substantially mitigated if the Supreme Court were simply to clarify that likelihood of confusion is a question of law based on underlying facts. Indeed, in Jack Daniel’s, the Supreme Court did note that early dismissal for no likelihood of confusion as a matter of law could sometimes be appropriate, though the Court did not clarify whether the ultimate issue is appropriately considered one of law or fact.198

C. Rogers Can Still Potentially Apply to Some NFTs

It is not at all clear how one is supposed to know when the plaintiff’s mark is being used by the defendant “as a mark” such that Rogers categorically does not apply in light of Jack Daniel’s. In stating that Rogers applies “only to cases involving ‘non-trademark uses’—or otherwise said, cases in which 'the defendant has used the mark’ at issue in a ‘non-source-identifying way,’” the Court cited to and quoted a law review article by Stacey Dogan and Mark Lemley.199 That article itself recognized that “[s]ometimes it is impossible to determine whether a defendant is using a mark to indicate its products’ source or sponsorship without resorting to the type of consumer-perception analysis that sits at the core of the likelihood-of-confusion test.”200

The Court pointed to Mattel v. MCA as an example where in the song “Barbie Girl,” the Barbie name was not used as a source identifier, so Rogers properly applied.201 But Barbie was in the title of the song, which seems like it could conceivably suggest a connection to the Barbie brand. Indeed, a recent film entitled Barbie was sponsored and produced in part by Mattel Films.202 The film is about the Barbie characters, so in that case, it does seem that the movie title is being used “as a mark.” It does not seem clear how the use of “Bad Spaniels” on a dog toy is being used “as a mark” any more than “Barbie” in the title of the song “Barbie Girl.”

The Court explained, quoting from the Mattel opinion, that “the band’s use of the Barbie name was not as a source identifier: The use did not speak to the song’s origin,” and that “a consumer would no more think the song was produced by Mattel than would, upon hearing Janis Joplin croon ‘Oh Lord, won’t you buy me a Mercedes Benz?, . . . suspect that she and the carmaker had entered into a joint venture.”203 But again, that does not seem so clear, some consumers upon seeing the title of the song might be confused into thinking that it was produced by Mattel. Indeed, that is precisely why the Rogers test is necessary, as Rogers itself involved a potentially confusing title, and the court felt that the likelihood of confusion inquiry alone was not sufficient to protect the First Amendment free expression interests at stake.

Perhaps one aspect is that even where there is a potential confusion based on a title alone, courts should consider whether consumers viewing the work as a whole would likely be confused. In Rogers and Mattel, the underlying works did not suggest a connection to Ginger Rogers or Barbie, whereas the Barbie movie as a whole does suggest such a connection. The Rogers doctrine may thus prevent the notion of “initial interest confusion” from being used to justify infringement based on the titles of artistic works alone.

In any event, saying that Rogers applies only when there is no confusion collapses Rogers into the likelihood of confusion analysis and makes Rogers irrelevant. Essentially then, one could read the Jack Daniel’s opinion as holding that Rogers applies only when it is not necessary, that is, only when consumers would not be confused into thinking that the use of the plaintiff’s mark indicates any sponsorship or affiliation from the plaintiff’s brand. The Court gave the use of a Louis Vuitton bag in a movie to convey something about one of the characters as an example of a case where the Louis Vuitton mark was not being used “as a mark” and thus Rogers could apply.204 But this would seem to be a situation where confusion as to source would be highly unlikely, so Rogers would be unnecessary.

On the other hand, given the circularity of the likelihood of confusion inquiry, if courts were to start to hold that the use of a product in a movie constitutes trademark infringement, then consumers may eventually begin to think that the use of a product in a movie does indicate some degree of sponsorship or affiliation. The Rogers doctrine then could function as a backstop to insulate such uses against infringement.205 It also can provide added protection for expressive uses of trademarks that do not indicate source, given the unpredictable and malleability of the likelihood of confusion inquiry. Given the difficulty of drawing clear lines in these areas, Rogers is likely best understood as a thumb on the scale against infringement when the allegedly infringing use of the mark is artistic or expressive and is not being used “as a mark.”206 Although this would only be applied in cases where the balance likely already tilts at least somewhat against confusion, this additional thumb on the scale could make it easier to dismiss the case early on, reducing litigation costs for artistic defendants and thus insulating protected expression against an in terrorem chilling effect.

Turning back to the MetaBirkin NFTs, one could argue that Birkin is not being used “as a mark” any more than Barbie in the song Barbie Girl. Indeed the district court in the Hermes case found that Rogers applied because the use of the Hermes marks in the MetaBirkins “did not function primarily as a source identifier.”207 However, that decision was made before the Jack Daniel’s decision, which stated that Rogers does not apply when the mark is being used “at least in part” for source identification.208 And one could certainly argue that Birkin is being used “as a mark” or for source identification in MetaBirkin at least as much as the Bad Spaniels dog toy is making use of the Jack Daniel’s brand for source identification.

As such, after Jack Daniel’s it seems that Rogers would probably not apply to the MetaBirkin NFTs.209 Given that the court found a likelihood of confusion, it may be difficult to argue that the name Birkin does not speak at least in part to the origin of the NFTs. And it would be even more difficult to argue that RR/BAYC is not using BAYC “as a mark.”

One can, however, imagine situations where Rogers could still apply to NFTs. For example, imagine a Bored Ape NFT where the Bored Ape is carrying a handbag that looks something like a Birkin bag, or perhaps even is identifiable as a Birkin bag. Or maybe the bored ape character is wearing Nike shoes. In these situations, the allegedly infringed trademark is playing a more minor role in the NFT and is not included in the title, such that it is arguably not being used “as a mark,” and Rogers at least arguably could apply. This would be more akin to the example the court gave of the use of a Louis Vuitton bag in one scene of a movie as a situation where Rogers could still properly have some application.

Thus, after Jack Daniel’s, Rogers can still arguably provide protection to some expressive NFTs, but probably only where the unauthorized use of the mark is less prominent than the uses at issue in the Hermes and Yuga Labs cases.

D. Courts Should Adopt a Flexible Approach to Rogers

The Jack Daniel’s Court purported not to be overruling or taking a position on the Rogers doctrine more generally, as far as how exactly it should apply when it does apply.210 Although the Court in Jack Daniel’s did not decide the issue of the correct approach to Rogers (if any), it did provide something of an invitation for courts to reconsider the issue.

As outlined in Part II, courts have split between essentially three different approaches to the Rogers doctrine. Courts led by the Second Circuit take a flexible approach, requiring a “particularly compelling” case for confusion when the allegedly infringing work is expressive. Courts led by the Ninth Circuit take a rigid approach, holding that an artistically relevant use of a mark is infringing only if it is “explicitly misleading.” And some courts, such as the Fourth Circuit, hold that Rogers is unnecessary because confusion is unlikely in the case of a true parody. The flexible approach is the most coherent and well developed, providing sufficient First Amendment protection while avoiding the need to draw arbitrary binary lines.

Although the likelihood of confusion factors can and should take into account whether the allegedly infringing work is expressive, the confusion inquiry alone does not seem well suited to fully account for some forms of expression such as parody.211 For example, although the similarity of the marks normally weighs in favor of confusion, in the case of parody, some degree of similarity is required to evoke the original, but the parodist should not necessarily be penalized for this. Similarly, generally the strength of the plaintiff’s mark weighs in favor of confusion, but in the case of a parody it may be the very thing that allows consumers to recognize the mark being ridiculed, and so its weight should likely be diminished in this context.

As such, the approach taken by the Fourth Circuit is not always adequate, and some added protection under Rogers may be appropriate. In other words, although courts can try to fit the likelihood of confusion factors into a parody analysis, this is to some extent like fitting a square peg in a round hole, so some additional doctrinal protection for expressive works such as parodies may be helpful.212

As far as the rigid approach, there is no good reason that trademark infringement in the case of an expressive use of a mark (now, when not used “as a mark”) should turn solely on whether the use is implicitly or explicitly misleading, and it is not even clear what that means. Even courts, such as those in the Ninth Circuit, purportedly using the “explicitly misleading” formulation do not limit the inquiry to that question and appear to mean “very misleading.”213 If courts mean something else when they say “explicitly misleading,” then they should say something else, in the name of judicial candor, to focus the inquiry appropriately, and to allow the law to develop in a more coherent and less confusing manner.214

As shown in Part II, the “explicitly misleading” approach essentially results from a misreading and overextension of one particular sentence from the Rogers decision. That sentence merely suggested how the more general balancing of free expression against consumer confusion would “normally” come out in the situation of artistic titles using celebrity names. It is worth emphasizing that even the judge that created the “explicitly misleading” formulation appears to have repudiated it.215 It is difficult to say whether the Ninth Circuit’s approach is significantly more protective of free expression as compared to the Second Circuit’s approach. What seems clear though is that the Ninth Circuit’s approach is less coherent and does not focus as directly on the inquiry of the actual values at stake.

Although the Supreme Court did not end up addressing the appropriate Rogers standard in Jack Daniel’s, a colloquy with Justice Gorsuch during oral argument showed some appropriate skepticism as to the “explicitly misleading” test.216

Justice Gorsuch also wrote a concurrence in Jack Daniel’s cautioning courts to “handle [Rogers] with care” noting that “it is not entirely clear where the Rogers test comes from” and asking “is it commanded by the First Amendment, or is it merely gloss on the Lanham Act, perhaps inspired by constitutional-avoidance doctrine?”217 The Rogers court did seem to be purportedly applying constitutional avoidance.218 But the court did not discuss whether its saving interpretation was reasonable or fairly possible as required by that doctrine.219 Another advantage of the Second Circuit’s approach over that of the Ninth Circuit’s is that it is more closely linked to the text of the Lanham Act, for although the words “explicitly misleading” do not appear in the relevant statutory text, the words “likely to cause confusion” do appear.220 But the Act does not state how likely the confusion must be, so it seems reasonable to interpret the Act to require a more compelling case for confusion where the allegedly infringing work implicates First Amendment concerns.

Another problem with the Ninth Circuit’s approach is that it leads to an arbitrary binary distinction between expressive works entitled to strong protection, and other works entitled to no protection.221 If there is one thing that the Supreme Court made clear in Jack Daniel’s, it is that this line will not be drawn so as to give full First Amendment protection to anything with a mere modicum of expressiveness.

The Second Circuit’s flexible approach helpfully allows for more of a sliding scale, analyzing the degree to which the likelihood of confusion needs to be “particularly compelling” in relation to the weight of the First Amendment interest on the other side of the balance.222 The stronger the case for First Amendment protection, the more compelling the case for confusion must be to outweigh it. True parody or creative criticism should require a very compelling case for confusion given the strong First Amendment interests at play.223

On the other side of the balance, a successful parody is not generally confused with the original but rather is recognized to be criticizing the original.224 So while a plaintiff targeting a transformative parody not only must meet a higher bar in showing confusion, the plaintiff also tends to have a weaker case for confusion.225

This article has argued that likelihood of confusion generally should be considered ultimately an issue of law,226 and the balancing of likelihood of confusion against the free expression interests for an allegedly infringing expressive work should a fortiori be considered a legal issue. The Court in Hermes arguably erred by treating the issue as too much of a factual one, essentially punting to the jury not only on weighing the various likelihood of confusion factors but also even on the Rogers balancing of confusion against free expression.227

The primary argument in favor of the Ninth Circuit’s rigid and somewhat incoherent explicit misleadingness approach is that it is necessary to prevent properly protected speech from being chilled by the high cost of litigating the amorphous and unpredictable issue of likelihood of confusion. But this concern would be mitigated by treating the likelihood of confusion question generally as one of law as argued above, or at least doing so in cases where Rogers applies, and the likelihood of confusion must be balanced against First Amendment free expression interests. Moreover, in light of Jack Daniel’s, Rogers will likely only apply in cases where confusion is rather unlikely, so courts should not have much trouble in such cases deciding well before trial that as a matter of law there is no sufficiently compelling case for confusion to outweigh the First Amendment free expression interests at stake.

E. First Sale Doctrine Should Generally Not Protect Unauthorized NFTs

The first sale doctrine is closely tied to likelihood of confusion, thought it can potentially be useful from a procedural perspective in straightforward cases, enabling quick dismissal without expensive litigation on the likelihood of confusion factors.228 The use of NFTs to track physical products, however, does not present such a straightforward case for the first sale doctrine. At the very least, an NFT digital version of a purchased physical product is likely “materially altered,” (at least by being converted into digital form) and thus the first sale doctrine should not apply.229 An NFT representing a pair of sneakers is simply not the same thing as a physical pair of sneakers, just as an NFT depicting a Birkin bag is not the same thing as a physical Birkin bag, and a painted image of a pipe is not the same thing as a physical pipe.230

In general, NFTs should be considered separate products from their physical counterparts (when there is such a counterpart), and thus buying a product should not necessarily give the buyer a right to create an NFT of that product. That is certainly not to say that there is necessarily trademark infringement, but whether there is should depend (as usual) on whether consumers are likely to be confused.

The factor of degree of similarity between the products or goods presents an interesting question when we are faced with an NFT with a physical counterpart. One might expect an NFT of a Nike shoe to have been made by Nike more than one might expect an Andy Warhol style painting of a Nike to have been made by Nike; although most established brands do not generally make paintings of their products, many are starting to make NFTs of their products.231 So perhaps an NFT should be considered something like a t-shirt, a common place for brands to advertise and extend their reach into the digital world. But the more the NFT picture is stylized or altered from its original, as in the case of the fuzzy Birkin bags,232 the more this factor should tend to weigh against confusion.

Of course, similarity of the goods is just one factor. Even if the NFT contains an unvarnished picture of the product, courts should remain open to the possibility that where an NFT truly is being used as just a technological tool to track a physical product, and this is made completely clear to consumers, confusion may not be likely. In such a situation, noninfringement of the NFTs could be found based on no likelihood of confusion, whereas resale of the physical goods would be protected by the first sale doctrine.

In discussing the Nike v. StockX case, Mark McKenna has recently seemingly questioned whether NFTs should be considered separate products from the underlying shoes in this context, and argued that NFTs should generally not be protected by trademark law, in part because marks on NFTs don’t convey information about quality in the sense of “durability or comfort, suitability for running, or anything analogous.”233 He recognizes the counterargument that consumers might still care about the origin of digital goods such as NFTs “just because the care about ‘authenticity’ as such—not because those digital shoes have characteristics as shoes, or even because the digital Nike shoes from Nike will ‘work’ differently than other digital Nike shoes.”234 But he rejects this counterargument, in part because the information about source can be recorded in the blockchain, that is, “[i]f you want to know if my avatar’s digital shoes really came from Nike, that information can be carried in the token.”235

It is certainly true that information about source can be carried in an NFT token, but the question is whether a typical consumer would be aware of this information. To the extent that consumers are made aware of this information at the point of sale, that would diminish the likelihood of confusion and thus the likelihood of trademark liability. If consumers are not adequately made aware, confusion is still a possibility. In any event, the fact that the blockchain may provide a mechanism for avoiding confusion is no reason to use the first sale doctrine to categorically exclude NFTs from the likelihood of confusion analysis.

Professor McKenna also argues that if one were to “strip away the NFTs from the StockX case, [then] it’s a first sale case where the question is whether the fact that StockX’s authentication process is imperfect makes it a counterfeiter.”236 That does not seem correct, in part because a painting, photograph, or digital image containing a Nike symbol is not the same thing as a Nike itself; it is a material alteration of the actual Nike shoe and as such should not be covered by the first sale doctrine. Buying a Nike shoe does not give one the right under the first sale doctrine to start selling artistic paintings or photographs of the Nike, and the fact that the image is linked to or sold via an NFT should not change that outcome. The artist’s sale of the image may be protected under Rogers (though that is far from clear after Jack Daniel’s), or there may be no likelihood of confusion, but regardless, there is not a viable first sale defense.

One might also attempt to analogize the Nike v. StockX situation to a coat check where one receives merely a plain claim ticket. If one were then to sell the claim ticket including the rights to the underlying coat, would that be covered by the first sale doctrine? The transfer of the coat itself would certainly be covered, as would the transfer of the underlying Nike’s. With respect to the claim ticket, there would be no reason to think that it was made or sponsored by the coat manufacturer, so there would be no likelihood of confusion and no trademark infringement. That would be the key distinction from the StockX case, where the NFT contains the Nike trademark and thus at least arguably creates confusion. But the coat check ticket is still a separate item from the coat itself, so if there were possible confusion as to whether it was made by the coat manufacturer, the seller of the ticket should not necessarily be protected from a trademark claim merely because they lawfully purchased the coat.

Another difference is that unlike plain claim tickets NFTs are a new and separate category of products or investments that are usually not even linked to any physical item. The separateness of the NFTs from the Nikes is also evidenced by the fact that the NFTs seem to sell for prices greater than and uncoupled from the value of the Nikes themselves. Thus, an NFT would seem to be a separate product from, or at least a material alteration of, an associated physical item, so a trademark claim on the NFT should turn on likelihood of confusion rather than a first sale defense founded in the lawful purchase of any associated physical item.

Conclusion

The general confusion surrounding the likelihood of confusion inquiry in trademark law is exacerbated by the general confusion surrounding NFTs. Given the degree of circularity inherent in the likelihood of confusion analysis and the relative novelty of the context, courts have an important opportunity to shape how the confusion analysis will apply to NFTs.

When applying likelihood of confusion to NFTs, courts should keep in mind that the value of NFTs is largely based on prestige or clout, and thus courts should ask whether any alleged confusion would increase the amount that buyers are willing to pay for the NFTs. For example, if the NFT is making unauthorized use of a luxury brand’s trademark, this could well increase the amount that buyers would be willing to pay for the NFT, as the prestige of the luxury brand would likely add to the clout and thus the value of the NFT. This consideration should factor into the weight given to any evidence of actual confusion, such as survey evidence. Likelihood of confusion should be considered a question of law based on underlying facts, so ultimately it should be up to the court to weigh and balance the various factors.

With respect to parody and the First Amendment, courts should move in the direction of a flexible approach to Rogers, such as that taken by the Second Circuit, where a more compelling case for confusion is required to outweigh the free expression interests at stake. This balancing should also (and a fortiori) be considered a legal matter for the court. The balancing can still apply even after Jack Daniel’s where an NFT makes a less prominent or more minor parodic use of another’s trademark without using aspects of that trademark as its own trademark or to indicate source. Evidence of confusion as to whether a parody was authorized by the brand being ridiculed should be given less weight if it does not appear that such confusion is likely to alter consumer purchasing decisions.

The first sale doctrine should generally not protect the use of NFTs to track physical products, as the NFT should be considered a separate product or at least a material alteration. But where it is made clear to consumers that the NFT is merely being used as a technological tool to track an underlying physical product and was not produced by the same brand that made the physical product, confusion may be unlikely.