In brief. Trademark law was built, over the better part of two centuries, to keep buyers from being fooled about who made the soap, the whiskey, or the handbag. Now it is being asked to govern handbags that exist only as JPEGs on a blockchain. The Hermès International v. Rothschild "MetaBirkins" litigation supplied the first full-dress doctrinal answer—and its still-pending Second Circuit appeal will sharpen it. The decisive shift, though, came from the Supreme Court: Jack Daniel's Properties, Inc. v. VIP Products LLC, 599 U.S. 140 (2023), withdrew the Rogers v. Grimaldi free-speech shield wherever a mark is used as a source identifier, which is exactly how most infringing NFT projects use the marks they borrow. This article traces that doctrine end to end—use in commerce, likelihood of confusion, dilution, the Rogers test after Jack Daniel's, and the Nice Classification's scramble to catalogue goods that have no weight. A companion article covers what to do about all this. This one is about what the law actually says. None of it is legal advice.

In December 2021, at Art Basel Miami, an artist named Mason Rothschild unveiled one hundred non-fungible tokens depicting fuzzy, fur-covered handbags. He called them "MetaBirkins." Each was, unmistakably, a Hermès Birkin—the silhouette is one of the most recognizable shapes in luxury—rendered in pixel-perfect pelt. They sold for around $450 apiece and the secondary market promptly bid some of them into five figures. Rothschild called the project a "playful abstraction," a commentary on fashion's appetite for animal skins. Hermès called it theft and sued.

The lawsuit that followed did something no congressional hearing or law-review symposium had managed to do. It forced a federal jury to decide, in concrete dollars, whether the word and design that sell a $20,000 leather bag also reach a $450 picture of a fake one. The answer—yes, on these facts—has become the load-bearing precedent for an entire emerging economy. Forecasts for immersive commerce run into the hundreds of billions of dollars; hundreds of millions of users already roam platforms where avatars wear branded sneakers, carry branded bags, and drive branded cars, none of which exist in any warehouse. Every brand counsel now confronts the same uncomfortable questions Hermès did. Can a registration covering physical handbags reach a virtual one strapped to an avatar? Does the international classification system even have a box for "downloadable digital handbag"? And how do you prove that consumers were confused by a project minted by a pseudonymous wallet on a decentralized ledger, sold to buyers scattered across every jurisdiction on earth at once?

The reassuring news, which we will earn over the next several thousand words, is that trademark law has not been blown up by the metaverse. The touchstone is still likelihood of confusion. Famous marks still get broad protection. Commercial intent still corrodes a free-speech defense. What has changed is the terrain on which those familiar principles operate—and a handful of those principles, particularly the relationship between a physical product and its virtual twin and the boundary of the Rogers defense, are still being worked out in real time.

To keep the doctrine concrete, we will run a hypothetical alongside the real cases. Meet Cordova Maison (hypothetical), a storied luxury accessories house with famous word marks, a distinctive trade-dress silhouette, and a portfolio thick with registrations for leather goods, jewelry, and ready-to-wear—but, until last quarter, not a single registration covering anything virtual. Cordova faces the Hermès predicament from both ends: an unauthorized "CordovaVerse" NFT drop is trading on its name on a secondary marketplace, and Cordova's own board is debating whether to open a flagship store inside a metaverse platform. Cordova's lawyers need to know what the law is before they decide what to do. So do we.

This article is the doctrine-and-landscape map. For the operational playbook—portfolio architecture, watch services, decentralized enforcement, takedowns, and licensing terms—see our companion piece, Trademark challenges in the metaverse: brand-protection strategies for digital-first commerce. The two are meant to be read together; this one supplies the why, that one supplies the how.

First Principles: What a Trademark Protects, and Why That Matters Online

Before we put an avatar in a virtual storefront, it helps to remember what a trademark is for. A trademark is not a property right in a word or a shape for its own sake. It is a right to prevent others from using a mark in a way that confuses consumers about the source of goods or services. The Lanham Act protects registered marks against infringement under Section 32 (15 U.S.C. § 1114) and protects registered and unregistered marks against infringement and false designation of origin under Section 43(a) (15 U.S.C. § 1125(a)). Famous marks get an additional, distinct protection against dilution under Section 43(c) (15 U.S.C. § 1125(c)), which we reach below. If you want the ground-floor version of all this, our trademark basics primer lays it out, and the difference between copyright, trademark, and patent explains where trademark fits among the IP regimes.

Two features of that definition matter enormously in virtual contexts.

First, trademark rights are tethered to use in commerce. The Lanham Act defines "use in commerce" as the bona fide use of a mark in the ordinary course of trade—affixed to goods sold or transported, or displayed in the sale or advertising of services (15 U.S.C. § 1127). For most of trademark history, "use" meant something physical: a label sewn into a collar, a logo stamped on a bottle. Minting an NFT and selling it for cryptocurrency is, just as plainly, a commercial transaction—buyers pay, sellers profit, the mark appears in the listing and the artwork. The MetaBirkins jury had no trouble treating the sale of NFTs as commercial activity within the Lanham Act's reach. The harder cases live at the margins: an avatar "skin" given away free to drive engagement, a fan-made digital wearable nobody charges for, a logo that materializes inside a game world the user never paid to enter. The further a use drifts from a paid transaction, the weaker the "use in commerce" hook—and the more a brand owner may have to lean on dilution (which has its own commercial-use requirement) or simply on platform terms of service rather than the Lanham Act.

Second, trademark rights are defined by goods and services, not by ideas. When the USPTO registers CORDOVA for "handbags, namely leather purses" in Class 18, the registration's scope is anchored to that description. The whole metaverse classification debate, which occupies a long section below, is a fight about whether a registration moored to physical handbags floats over to virtual ones—or whether a brand must file separately for "downloadable virtual handbags" to be safe. The answer turns out to depend heavily on where you ask the question. For the basics of how the USPTO sorts goods, see our explainer on USPTO trademark classes.

Hold onto both features. "Use in commerce" and "which goods and services" are the two pressure points where virtual goods stress the doctrine the most.

The MetaBirkins Case: The Doctrinal Anchor

The litigation between Hermès and Mason Rothschild (legal name Sonny Estival) remains the most significant judicial treatment of trademark rights in NFTs anywhere in the United States. It is worth understanding in detail, because nearly every metaverse trademark argument now either cites it or distinguishes it.

Rothschild had a track record. In May 2021 he sold a "Baby Birkin" NFT—a digital image of a Birkin bag with a gestating fetus visible inside it—for roughly $23,500. Emboldened, he produced the MetaBirkins collection: one hundred NFTs of fur-covered Birkins, branded with the "MetaBirkins" name and a matching website at metabirkins.com, debuting at Art Basel at about $450 each. Hermès sent a cease-and-desist in December 2021. Rothschild ignored it, publicly framing himself as an artist standing up to a corporate bully. In January 2022, Hermès sued in the Southern District of New York for trademark infringement, dilution, and cybersquatting, arguing that Rothschild was, in its words, "stealing the goodwill in Hermès' famous intellectual property." Rothschild answered that the MetaBirkins were protected artistic expression—a comment on luxury, cruelty, and the fashion industry's relationship with fur.

The case lived and died on Rogers v. Grimaldi, 875 F.2d 994 (2d Cir. 1989), the Second Circuit's framework for trademarks used in expressive works. Under Rogers, the use of another's mark in an expressive work is protected unless the use (a) has no artistic relevance to the underlying work, or (b) explicitly misleads as to the source or content of the work. Judge Jed Rakoff denied cross-motions for summary judgment in a closely watched 2022 opinion. He held that Rogers applied—NFTs can be expressive works, so Rothschild's project was not categorically outside the First Amendment's shade—but that a reasonable jury could find Rothschild's use explicitly misleading. That sent the case to trial on the misleadingness prong, with intent and actual confusion squarely in play.

The trial, in late January and early February 2023, was an object lesson in how evidence wins trademark cases. Hermès did not argue abstractions about luxury and code. It put Rothschild's own words in front of the jury: text messages and social-media posts in which he described the project as "a gold mine," mused about Birkin's "insane" markups, and talked about wanting to "make big money." It introduced a consumer survey showing roughly an 18.7% net confusion rate—a fifth of relevant consumers thought Hermès was behind, or had approved, the MetaBirkins. And it showed that the fashion and crypto press had repeatedly reported the MetaBirkins as an official Hermès venture, the kind of actual marketplace confusion that survey numbers only estimate. On February 8, 2023, the jury found for Hermès across the board—infringement, dilution, and cybersquatting—rejected the First Amendment defense, and awarded $133,000 in damages. In June 2023, Judge Rakoff entered a permanent injunction barring Rothschild from marketing the MetaBirkins. His framing in the post-trial rulings has been quoted ever since: Rothschild would have been protected, he observed, had the jury found "even a modest semblance of artistic expression," but the jury instead concluded that "Rothschild was simply a swindler."

Three lessons crystallize out of the verdict, and they generalize well beyond Hermès.

First, the medium confers no immunity. "It's an NFT" is not a defense, any more than "it's a painting" or "it's a T-shirt" would be. The same Lanham Act applies; the inquiry is the same; the blockchain is just a distribution channel. That is the single most important takeaway for Cordova's counsel staring down "CordovaVerse."

Second, famous marks reach across categories. Hermès did not, at the relevant time, hold registrations specifically covering virtual goods or NFTs. It won anyway, on the strength of the BIRKIN mark's fame and the relatedness of the parties' goods. A famous mark casts a long shadow. (That said—see the classification section—do not read this as permission to skip filing for virtual goods. Hermès won the hard way, on a developed factual record; a thinner record could break the other way.)

Third, the verdict is fact-bound, not a rule of automatic liability. Hermès prevailed because of unusually damning intent evidence and unusually clear confusion evidence. Strip those away—give the artist a genuinely transformative work, a prominent disclaimer, no "gold mine" texts, and weak survey numbers—and the outcome is far less certain. MetaBirkins is a ceiling on what artists can get away with, not a floor that swallows all expressive use. For Cordova, the operative lesson is evidentiary: if you are going to enforce, build the confusion-and-intent record early and deliberately, the same documentary discipline we describe in the context of trade secret protection programs. The facts that proved decisive against Rothschild were facts somebody had to find, preserve, and present.

The appeal, and why it still matters

Rothschild appealed. The Second Circuit heard oral argument on October 23, 2024, and the reports from the bench described a divided panel—two judges seemingly sympathetic to Hermès, one to Rothschild, with Senior Judge Pierre Leval reportedly suggesting the district court may have instructed the jury "improperly" on the role of intent under Rogers. As of mid-2026, the court has not issued its decision. It remains the most closely watched open question in metaverse trademark law.

The stakes of the appeal are doctrinal, not just personal to one artist. A ruling for Rothschild that faults the jury instructions would reaffirm meaningful breathing room for artists who genuinely invoke famous marks in their work, and would discipline how district courts charge juries on intent. A ruling for Hermès would entrench aggressive enforcement against source-identifying NFT uses. Either way, the larger framework has already shifted under the case's feet—because while the appeal sat pending, the Supreme Court rewrote the rules the appeal depends on.

Jack Daniel's Changes the Backdrop

In Jack Daniel's Properties, Inc. v. VIP Products LLC, 599 U.S. 140 (2023), the Supreme Court confronted a squeaky dog toy. VIP Products sold a rubber chew toy shaped like a Jack Daniel's bottle, relabeled with bathroom humor ("Bad Spaniels," "The Old No. 2 on Your Tennessee Carpet"). The Ninth Circuit had applied Rogers, treating the toy as an expressive work and clearing it. The Supreme Court reversed—and in doing so, narrowed Rogers in a way that reverberates straight into the metaverse.

The Court's holding is narrow on its face and enormous in effect. Rogers, the Court held, does not apply at all when the accused infringer uses the challenged mark "as a designation of source for the [defendant's] own goods"—that is, as a trademark. In that posture, the ordinary multifactor likelihood-of-confusion test governs, and the expressive content of the work, however clever, does not buy a free pass. Justice Kagan, writing for a unanimous Court, was explicit that Rogers is at most a "cabined" doctrine for cases where a mark is used not to identify source but as part of an expressive message. The toy used "Bad Spaniels" to brand and sell toys; therefore Rogers dropped out and confusion analysis took over. (The Court separately rejected VIP's dilution fair-use argument, holding that the noncommercial-use and parody exclusions in the dilution statute do not apply when the parody is itself used as a source identifier—a point we return to under dilution.)

Map that onto NFTs. Most infringing virtual-goods projects do exactly what VIP did: they take a famous mark and use it as the name and brand of the thing they are selling. "MetaBirkins" was the brand of Rothschild's collection, splashed across the website and the social accounts and the listings. Under Jack Daniel's, that use looks far less like an expressive reference and far more like ordinary source-identifying use—meaning Rogers may not apply at all, and Hermès need only prove likelihood of confusion. The very framework Rothschild relied on at trial may have shrunk beneath him on appeal. That is the central appellate question now: not whether Rothschild satisfied Rogers, but whether Rogers was ever the right test for a mark used as a brand name.

The line Jack Daniel's draws is the line every creator and every brand counsel now has to internalize: it is the difference between using a mark as content within a work and using it as the brand under which the work is sold. A painting that depicts a Birkin and is sold under the artist's own name is using the mark as content. A collection called "MetaBirkins," sold at metabirkins.com, is using it as a brand. The first keeps its Rogers argument; the second, after Jack Daniel's, very likely loses it.

Likelihood of Confusion in Virtual Environments

Strip away the special defenses and infringement always comes back to one question: is consumers' confusion as to source likely? Every circuit answers it with a multifactor balancing test—the Second Circuit's Polaroid factors (Polaroid Corp. v. Polarad Electronics Corp., 287 F.2d 492 (2d Cir. 1961)), the Ninth Circuit's Sleekcraft factors, the Fifth Circuit's "digits of confusion," and so on. The factors vary in number and labeling but cluster around the same ideas: strength of the senior mark, similarity of the marks, proximity or relatedness of the goods, evidence of actual confusion, the defendant's intent, the marketing channels used, and the sophistication of the relevant consumers. (These factors are flexible and non-exhaustive; courts are free to weigh others, as the Fifth Circuit reminds in National Business Forms & Printing, Inc. v. Ford Motor Co., 671 F.3d 526, 532 (5th Cir. 2012).) For a deeper treatment of how those factors operate, see our guide to navigating trademark confusion and, for the Second Circuit specifically, the Polaroid factors on summary judgment.

The doctrine is familiar. Its application to virtual goods bends in instructive ways.

Proximity (relatedness) of the goods is the factor most transformed. Traditional analysis treats handbags and software as commercially remote—different shelves, different buyers, different worlds. But a virtual handbag is not generic software; it exists precisely to evoke the physical handbag it copies. Its entire commercial appeal is that it looks like, and signals the status of, the real thing. So the psychological and commercial distance between a Birkin and a MetaBirkin is far smaller than the classification distance (Class 18 versus Class 9) would suggest. Courts that take the consumer's perspective—rather than the registrar's filing cabinet—tend to find virtual and physical luxury goods closely related, and that intuition drove both the MetaBirkins verdict and the Chinese "G.PATTON" decision discussed below.

Bridging the gap—the likelihood that the senior user will expand into the junior user's market—also shifts. Once upon a time a luxury house entering the video-game-skin business would have been a surprise. Today it is the expected next move. Nike, Gucci, Tiffany, Adidas, and dozens of others have launched virtual goods; expansion into the metaverse is now foreseeable as a matter of judicial notice. That makes the gap shorter and the confusion more plausible.

Consumer sophistication cuts in genuinely contradictory directions, and counsel should resist a glib answer. On one hand, some virtual goods command prices rivaling physical luxury items, and a buyer dropping four figures on an NFT may research carefully. On the other, metaverse demographics can skew young and impulsive, secondary marketplaces are dense with knockoffs and look-alikes, and the speed of a "mint now" drop discourages diligence. The same factor can favor either side depending on the platform and the price point.

Actual confusion remains the most persuasive evidence a plaintiff can muster, and—crucially—it is available in virtual contexts, sometimes more abundantly than offline. Hermès did not have to speculate; it had press outlets that had actually reported the MetaBirkins as official, plus survey data. Social media, marketplace reviews, and comment threads leave a confusion paper trail that a diligent brand owner can capture. (Capturing it properly is its own discipline; see our guide to authenticating website screenshots as evidence, because a screenshot that cannot be authenticated is worth nothing at trial.)

Intent does enormous work, as MetaBirkins showed. A defendant who deliberately chooses a famous mark to free-ride on its goodwill invites an inference of likely confusion—and, as a practical matter, juries punish bad actors. The "gold mine" texts were not technically an element of any claim, but they colored every factor in the case.

The bottom line: the toolkit is familiar, the readings shift, and the cases will accrete factor by factor. Counsel evaluating a virtual-goods dispute should run the full multifactor analysis precisely as in a physical-goods case—and then ask, for each factor, how the virtual context nudges it. The honest answer in 2026 is that relatedness and bridging-the-gap usually tilt toward the brand owner, while sophistication and the thinness of established case law leave genuine room to argue.

Dilution: The Famous Mark's Extra Armor

Infringement protects consumers from confusion. Dilution protects marks—specifically famous marks—from the slow erosion of their distinctiveness, even where no one is confused at all. As the Fourth Circuit put it in Rosetta Stone Ltd. v. Google, Inc., 676 F.3d 144, 167 (4th Cir. 2012), dilution targets the "whittling away" of a famous mark's unique selling power. Because the metaverse is, at the moment, disproportionately populated by the world's most famous brands—the ones with the resources and the incentive to plant flags early—dilution is a far more important weapon here than in run-of-the-mill trademark disputes.

The Trademark Dilution Revision Act of 2006 (TDRA), codified at 15 U.S.C. § 1125(c), sets the elements. A plaintiff must show that (1) it owns a mark that is both distinctive (inherently or through acquired distinctiveness) and famous; (2) the defendant used a mark in U.S. commerce after the plaintiff's mark became famous; and (3) the defendant's use is likely to cause dilution by blurring or tarnishment. Three features of that test matter intensely online.

Fame is a high bar, and it is the bar most often fatal to a dilution claim. "Famous" under the TDRA means widely recognized by the general consuming public of the United States as a designation of source—not merely famous within a niche. That is a demanding, nationwide standard; "niche fame" does not qualify. BIRKIN clears it. A boutique label known only to streetwear insiders may not. For Cordova, this is the threshold question its lawyers must answer honestly before pleading dilution: is the brand famous to the general public, or only to luxury consumers? (For how courts assess fame, our library draws on the standard cases; the leading Federal Circuit articulation is Coach Services, Inc. v. Triumph Learning LLC, 668 F.3d 1356 (Fed. Cir. 2012).)

Dilution does not require confusion, competition, or economic injury (15 U.S.C. § 1125(c)(1)). That is its power. A virtual-goods project might be cleverly enough disclaimed to escape an infringement finding—no one is confused—yet still blur the famous mark by making it less uniquely associated with its owner. Blurring is an association arising from similarity that impairs the famous mark's distinctiveness; courts weigh six non-exhaustive statutory factors (similarity, the senior mark's distinctiveness, the senior owner's substantially exclusive use, the mark's recognition, the defendant's intent to associate, and any actual association) (15 U.S.C. § 1125(c)(2)(B)). Tarnishment is harm to the famous mark's reputation by unsavory association (15 U.S.C. § 1125(c)(2)(C))—think a luxury mark slapped onto crude, violent, or scammy virtual goods. Both theories travel well to NFTs: the entire point of an unauthorized "CordovaVerse" drop is to summon the Cordova mark in the buyer's mind, which is the very association blurring forbids.

But dilution has real escape hatches, and Jack Daniel's reshaped them. The TDRA excludes from liability any fair use of a famous mark other than as a designation of source—including parody, criticism, and comment—plus any noncommercial use and all news reporting (15 U.S.C. § 1125(c)(3)). For years, parodists pointed to that exclusion as a near-automatic shield; the canonical example is Louis Vuitton Malletier, S.A. v. My Other Bag, Inc., 156 F. Supp. 3d 425 (S.D.N.Y. 2016), where a canvas tote joking "My Other Bag…" survived because it parodied, rather than branded itself with, the Louis Vuitton look. Jack Daniel's slammed that hatch partly shut: the Court held that the noncommercial-use and parody exclusions do not apply when the parody is itself used as a source identifier for the defendant's goods. A dog toy that brands itself "Bad Spaniels" cannot hide behind the parody exclusion, because it is using the joke as its trademark. The same logic dooms an NFT project that uses a famous mark as its own brand: the parody exclusion evaporates the moment the borrowed mark becomes the borrower's source identifier. For the broader doctrinal architecture of dilution, blurring, tarnishment, and these defenses, see our overview of trademark infringement and related rights.

The Nice Classification and Virtual Goods: A Framework Mid-Renovation

If likelihood of confusion is where virtual-goods disputes are litigated, classification is where they are prevented—or lost before they begin. The Nice Classification, the international system that sorts goods and services into 45 classes, was adopted in 1957 for a world of tangible products and physically rendered services. It did not contemplate a handbag with no leather, a sneaker with no rubber, a car with no engine. The result has been a worldwide scramble to figure out which box a weightless good belongs in—and, far more contentiously, whether owning the physical box also gets you the virtual one.

The classification picture has moved quickly:

  • Nice 12th Edition (in force January 1, 2023) first explicitly recognized NFT-related goods, adding "downloadable digital files authenticated by non-fungible tokens" to Class 9, alongside blockchain and crypto terminology across Classes 9, 41, and 42.
  • Nice 13th Edition (2024) refined the logic. The settled rule is now that an NFT is classified by the underlying good it authenticates, not as a free-floating "NFT." A digital image file authenticated by an NFT sits in Class 9; NFT-authenticated virtual clothing may implicate Class 25 reasoning by analogy; and virtual goods generally are treated as digital content in Class 9, with applicants required to specify the content with precision ("downloadable virtual goods, namely, virtual clothing," not the naked "virtual goods").

So the offices have converged on one point—specificity is mandatory—while splitting badly on the harder one: is a virtual good "related" to its physical twin for confusion purposes? This is the single most consequential open question in metaverse trademark practice, and the answer depends on your jurisdiction:

EUIPO. The European Union Intellectual Property Office requires that "virtual goods" be specified by nature and rejects bare claims as insufficiently clear and precise. Critically, the EUIPO has signaled it will refuse virtual-goods applications on likelihood-of-confusion grounds where they conflict with prior registrations for corresponding physical goods—implying a working presumption that the physical and virtual versions are related.

USPTO. The U.S. office, responding to a wave of filings led by Nike and other brands, asks applicants to indicate where virtual goods are used ("for use in online virtual worlds") and whether they are downloadable, and steers applicants toward specifications spread across Classes 9, 35, and 41. Like the EUIPO, the USPTO is prepared to find virtual goods confusingly close to physical counterparts—again presuming relatedness rather than distance.

KIPO. The Korean Intellectual Property Office charts the opposite course. Its guidance presumes that virtual goods and their real-world counterparts are dissimilar, given their different uses and consumption situations—while preserving broader, cross-category protection for well-known marks. Under KIPO's default, owning the physical registration does not automatically reach the virtual good; a separate filing is the prudent course.

China. A 2024 decision from the Hangzhou Intermediate People's Court bridged the divide in striking fashion. In the "G.PATTON" case, virtual vehicle skins inside an online game were found to infringe a trademark registered for physical vehicles in Class 12—even though the virtual skins lived in Class 9—on the reasoning that consumers could perceive a source connection between the two. China thus reached, through likelihood-of-confusion logic, a relatedness result that classification formalism would have forbidden.

EU policy. The European Parliament's Resolution on the policy implications of virtual worlds (October 17, 2024) confirmed that existing EU trademark law "fully applies" to the metaverse and that the Nice framework adequately encompasses virtual goods—an institutional reassurance that the law need not be reinvented, only applied.

The practical upshot for a brand like Cordova writes itself, and it cuts against complacency. Because relatedness cannot be assumed everywhere—KIPO presumes the opposite, and even the friendlier offices presume rather than guarantee it—the only defensible posture is belt-and-suspenders: maintain robust physical-goods registrations and file proactively and specifically for virtual goods and NFT-related services in every market that matters. The MetaBirkins win on cross-category fame is real, but it is a backstop, not a substitute for filing. For the mechanics of building that portfolio—and the strategic sequencing of which classes to file when—the companion strategy article is the right next read, and the fundamentals of clearance before you file are in our guide to conducting a comprehensive trademark clearance search.

Territoriality and Jurisdiction: A Borderless Market on a Bordered Doctrine

Trademark rights are territorial. A U.S. registration grants rights in the United States; it does not, by itself, reach Seoul or São Paulo. This first principle of trademark law collides head-on with a market that has no borders. An NFT minted by a pseudonymous wallet is, the instant it lists, available to buyers in every country at once. Whose law governs? Where is the "use in commerce"? Whom do you even sue?

The Lanham Act reaches some conduct abroad, but the Supreme Court has lately tightened that reach. In Abitron Austria GmbH v. Hetronic International, Inc., 600 U.S. 412 (2023)—decided, like Jack Daniel's, in the Court's consequential 2023 term—the Court held that the Lanham Act's infringement and false-designation provisions are not extraterritorial and apply only to claims where the infringing use in commerce is domestic. Foreign sales that never touched the United States fall outside the Act, even if the defendant is a U.S. company and even if some confusion eventually spills back home. For virtual goods, Abitron injects real uncertainty: when a sale happens on a global marketplace to a buyer of unknown location, pinning down a domestic use in commerce can be genuinely hard. Brand owners will often need to show U.S.-targeted marketing, U.S. buyers, or U.S.-based marketplace infrastructure to anchor a claim.

Personal jurisdiction over a pseudonymous, possibly foreign defendant is its own thicket—often the practical barrier that matters more than the doctrinal one. Tracing a wallet to a human, serving a defendant who may sit anywhere on earth, and enforcing a judgment against assets held in cryptocurrency are all real obstacles. When the defendant is in China—a frequent scenario given the volume of cross-border infringement—service itself becomes a project governed by the Hague Service Convention, which our practice has written about in detail in serving a China-based defendant under the Hague Service Convention. The doctrinal point for this article is simpler: territoriality strains in borderless commerce, Abitron has narrowed the U.S. statute's offshore reach, and brand owners should expect that a complete enforcement strategy will require coordinated, jurisdiction-by-jurisdiction filings rather than a single global lawsuit. The tactics of multi-forum enforcement belong to the companion strategy piece; the principle to carry away here is that there is no such thing as a single "metaverse jurisdiction," only the same patchwork of national systems, now stress-tested.

The Rogers Test After Jack Daniel's and MSCHF and Yuga Labs

We have met Rogers and Jack Daniel's separately; the appellate courts are now stitching them together, and the through-line is sharpening with each decision. The recalibrated rule, in a sentence: if the defendant uses the mark as a source identifier for its own goods, Rogers does not apply and ordinary confusion analysis governs; Rogers survives only for genuine expressive uses where the mark is content, not brand.

Two post-Jack Daniel's decisions show the rule biting in exactly the contexts that matter for digital goods.

In MSCHF Product Studio, Inc. v. Vans, Inc. (2d Cir. 2024), the Second Circuit applied Jack Daniel's to "Wavy Baby" sneakers that distorted Vans' Old Skool trade dress. The court held that Rogers did not protect the shoes, because MSCHF used Vans' trade dress in a source-identifying manner—the very use Jack Daniel's removes from the Rogers umbrella—and that the parody, such as it was, did not dispel the likelihood of confusion. A sneaker that is a parody of another sneaker, sold as footwear, is using the borrowed dress as a brand.

In Yuga Labs, Inc. v. Ripps, 144 F.4th 1137 (9th Cir. 2025), the Ninth Circuit addressed the "RR/BAYC" project, in which Ryder Ripps minted NFTs pointing to the same Bored Ape Yacht Club images, framed (Ripps argued) as protest art and commentary on Yuga. The court held that pointedly confusing NFT uses of another's marks fall outside Rogers protection—because the marks were deployed to identify and sell the defendants' own NFTs, not merely as expressive reference. Yuga is especially significant because it is squarely an NFT case from a different circuit, confirming that the MetaBirkins logic is not a Second Circuit idiosyncrasy. The Rogers-narrowing is national.

For digital creators, the synthesized lesson is stark but not nihilistic. Using an established mark as your project's name or brand carries serious risk; the Rogers shield you might once have invoked has thinned to the vanishing point for that use. But genuine commentary that references a mark inside a differently branded work—a piece of digital art that depicts a luxury bag, sold under the artist's own name, with the critique legible in the work itself—retains meaningful protection. The defensible posture is the one Rothschild conspicuously failed to occupy: make the artistic purpose evident in the work, not in a litigation affidavit; do not name your project after the mark; and remember that runaway commercial success (a "gold mine") can undercut the very expressive character you are claiming.

For brand owners, Jack Daniel's is a gift that should be used with judgment. It strengthens enforcement against commercial appropriation—source-identifying uses—while leaving purely artistic, non-source-identifying uses protected. The smart enforcement program distinguishes the two: pursue the brand-name knockoffs aggressively, and engage constructively (or simply leave alone) the genuine commentary, both because the latter will likely win and because heavy-handed enforcement against artists generates exactly the kind of bad press that follows a Streisand effect. Where the line between source-identifying and expressive use is genuinely contested, that is the battleground—and it is where survey evidence, disclaimers, and the defendant's own statements decide cases. The interplay with the First Amendment more broadly, including parody and nominative fair use, is the same balance courts strike in non-digital cases; our coverage of that balance lives across the trademark overview series.

A Worked Hypothetical: Cordova Confronts "CordovaVerse"

Let us run the doctrine through Cordova's facts (hypothetical), because abstractions are easy and applications are hard.

An anonymous wallet has launched "CordovaVerse," a collection of 500 NFTs depicting Cordova's signature quilted-flap handbag in neon colorways, sold on a major marketplace at 0.4 ETH each. The project's social account uses Cordova's exact word mark, the listings describe the items as "Cordova virtual bags," and a few crypto outlets have written it up as "Cordova's metaverse debut." Cordova has famous word marks and a registered trade-dress configuration for the quilted flap, registrations across Classes 18 and 25 for physical goods, and—as of last quarter—a pending Class 9 application for "downloadable virtual handbags." How does the doctrine sort this out?

Use in commerce? Easily satisfied. The NFTs are sold for cryptocurrency in paid transactions; the mark appears in the branding and the listings. This is commercial use, not a free fan project.

Likelihood of confusion? Strong for Cordova. The marks are identical (the project simply uses "Cordova"); the goods are highly related under the consumer-perspective relatedness analysis that carried MetaBirkins and G.PATTON; bridging the gap is a non-issue because Cordova has a pending virtual-goods application and the board is openly planning a metaverse store; actual confusion is documented in the press write-ups; and intent is suggested by the wallet's appropriation of the exact mark and the "Cordova's metaverse debut" framing. Sophistication is the only factor that meaningfully cuts the other way, and it is contested.

Does Rogers save the project? Almost certainly not. Under Jack Daniel's, MSCHF, and Yuga Labs, "CordovaVerse" uses the mark as the brand of its own goods—it is the project's name and source identifier—so Rogers likely does not even apply, and Cordova need only prove confusion, which it can. There is no legible artistic commentary here; this is a knockoff with a wallet, the "swindler" posture Judge Rakoff described.

Dilution? If Cordova's mark is famous to the general consuming public (a real question its lawyers must vet honestly), the blurring claim is powerful: the project's entire value proposition is to summon the Cordova mark in buyers' minds, which is precisely the association blurring forbids, and the parody exclusion is unavailable because the mark is used as a source identifier. If the colorways are crude or the project later turns out to be a scam, tarnishment is also in play.

Territoriality and jurisdiction? Here Cordova's confidence should drop. The wallet is pseudonymous and possibly offshore; Abitron means Cordova must anchor a domestic use in commerce (U.S. buyers, U.S.-targeted marketing, U.S. marketplace infrastructure); and even a favorable judgment may be hard to enforce against crypto-held assets and an unidentified defendant. The fastest practical relief may come not from a lawsuit at all but from a marketplace takedown under the platform's IP policy—which is exactly the kind of operational move the companion strategy article addresses, and which overlaps with the DMCA and platform-liability questions that govern user-generated infringement.

Notice the shape of the answer. On the merits, Cordova holds a commanding hand—the doctrine has matured to the point where a brand-name NFT knockoff is a losing proposition for the infringer. The hard part is no longer the law; it is reaching the defendant. That inversion—strong merits, weak enforcement reach—is the defining feature of metaverse trademark practice in 2026.

Adjacent Frontiers: AI-Generated Goods, Interoperability, and Publicity

Three developments sit just over the horizon and will shape the next phase of doctrine.

AI-generated virtual goods raise liability questions the cases are only beginning to answer. When a generative system produces an output incorporating a trademarked design—sometimes without any user prompting for a specific brand—who is liable: the user, the platform, the model's trainer? Courts have begun allowing claims against platforms whose systems can "recall and recreate" protected design elements, a problem that overlaps tightly with the ownership questions we examine in AI-generated inventions and the training-data disputes in copyright claims against generative AI. The metaverse and generative AI are converging fast; a virtual storefront stocked by a model trained on branded imagery is a lawsuit waiting for a plaintiff.

Interoperability and portable assets complicate enforcement at the structural level. The promise of the metaverse is that a wearable bought on one platform travels to another. But a takedown that scrubs an infringing item from Platform A may not reach the same token rendered on Platform B, and a brand owner chasing a portable good must play whack-a-mole across ecosystems that do not share a moderation system. The more interoperable the asset, the harder the enforcement.

Right of publicity rides alongside trademark whenever a virtual good invokes a person rather than a brand. An avatar styled after a celebrity, a digital double trained on a real likeness, an NFT trading on someone's persona—these implicate publicity rights that run parallel to, and sometimes overlap with, false-endorsement claims under Lanham Act § 43(a). The doctrine there is evolving just as fast; see our primers on right of publicity basics and on the right of publicity meeting digital doubles—deepfakes, AI avatars, and celebrity likeness. And where the virtual good is itself an NFT resold on a secondary market, the trademark questions intertwine with the royalty-and-smart-contract questions analyzed in NFT marketplaces and secondary-sale royalties.

Conclusion: The Law Held; the Terrain Moved

The metaverse promised, or threatened, to break trademark law. It has not. As MetaBirkins demonstrated and Jack Daniel's confirmed, the core principles translate cleanly to virtual goods when applied with care. Likelihood of confusion is still the touchstone. Famous marks still cast a long, cross-category shadow, and now wield dilution as extra armor. Commercial motivation still corrodes a free-speech defense. And source-identifying use—using a mark as a brand rather than as content—still forfeits the Rogers shield, which is precisely how the great majority of infringing NFT projects use the marks they take.

What changed is not the doctrine but the ground it stands on. Goods now exist without mass. Markets now span every jurisdiction simultaneously. Defendants now hide behind wallets. Classification systems built for leather and glass strain to catalogue code, and the offices disagree—sometimes sharply, as KIPO and the USPTO do—about whether owning the physical good gets you the virtual one. Those frictions are real, and the pending Second Circuit decision in Hermès v. Rothschild will smooth some of the Rogers edges. But the foundational architecture is sound.

For brand owners like Cordova, the landscape rewards the proactive over the reactive: file specifically for virtual goods and NFT services rather than relying on cross-category fame as a backstop; establish a virtual presence so that bridging-the-gap and relatedness tilt your way; and build the confusion-and-intent record early, because those facts decide cases. For digital creators, MetaBirkins draws a boundary without foreclosing all trademark-referencing art—provided the creative purpose is genuine and legible in the work, and the mark is not pressed into service as a brand. For practitioners, the comforting news is that the doctrines you already know—confusion, dilution, fair use, territoriality—are the doctrines that govern the metaverse, even as the contexts mutate around them.

The next move is operational. If this article answered what the law says, the question of what to do about it—portfolio strategy, monitoring, takedowns, licensing, decentralized enforcement—belongs to the companion brand-protection playbook. Read them as a pair.

This article is intended for informational purposes only and does not constitute legal advice. The MetaBirkins appeal and metaverse trademark law generally are evolving rapidly; readers should confirm the current posture of the Second Circuit appeal and any office practice before relying on this summary, and should consult qualified trademark counsel regarding specific circumstances. For assistance, please contact our intellectual property and technology practice.

Related Articles

Frequently Asked Questions

Can a trademark registration for physical goods stop someone from selling a virtual version? Sometimes—but do not count on it. In the United States, a famous mark can reach virtual goods through cross-category relatedness, which is how Hermès prevailed in MetaBirkins despite lacking virtual-goods registrations at the time. But relatedness is presumed in some offices (EUIPO, USPTO lean that way) and presumed against you in others (Korea's KIPO defaults to treating virtual and physical goods as dissimilar). The safe course is to register specifically for virtual goods and NFT-related services, with precise specifications, in every market that matters.

Is selling an NFT "use in commerce" for trademark purposes? Yes, when the NFT is sold for value in a real transaction and the mark appears in the branding, listing, or artwork—that is bona fide commercial use within the Lanham Act (15 U.S.C. § 1127). The closer question is free or giveaway virtual goods, where the "use in commerce" hook weakens and a brand owner may have to rely on dilution or platform terms of service instead.

Does the First Amendment protect an artist who uses a famous mark in an NFT? It can, but the protection narrowed sharply after Jack Daniel's Properties, Inc. v. VIP Products LLC, 599 U.S. 140 (2023). The Rogers v. Grimaldi expressive-works defense does not apply where the artist uses the mark as a source identifier—the brand or name of the goods being sold. It survives only for genuine expressive use where the mark is content within a differently branded work. Naming your project after a famous mark, as "MetaBirkins" did, is the use most likely to lose the defense.

What is the difference between trademark infringement and dilution in this context? Infringement protects consumers from confusion about source; you must prove likelihood of confusion. Dilution protects famous marks themselves from the erosion of their distinctiveness and requires no confusion at all—only fame, plus blurring or tarnishment (15 U.S.C. § 1125(c)). Dilution matters disproportionately in the metaverse because the early adopters are the world's most famous brands, and because it can reach cleverly disclaimed projects that escape an infringement finding.

Which Nice class do virtual goods and NFTs go in? Under the current Nice editions, virtual goods are generally treated as digital content in Class 9, and NFTs are classified by the underlying good they authenticate rather than as NFTs in the abstract. Offices require precise specifications—"downloadable virtual goods, namely, virtual clothing," not bare "virtual goods"—and brands often file across Classes 9, 35, and 41 to cover the goods, retail services, and online entertainment dimensions.

If the infringer is anonymous or overseas, can I still win? On the merits, very likely—the doctrine strongly favors brand owners against brand-name NFT knockoffs. The obstacle is reach. Abitron Austria GmbH v. Hetronic International, Inc., 600 U.S. 412 (2023), confirmed the Lanham Act applies only to domestic uses in commerce, so you must anchor a U.S. nexus; and identifying, serving, and collecting against a pseudonymous, possibly foreign defendant is hard. In practice, a marketplace takedown under the platform's IP policy is often the fastest meaningful relief, with litigation reserved for defendants you can actually find.

Selected Authorities

Cases: Hermès International v. Rothschild, No. 22-cv-384 (S.D.N.Y. 2023) (jury verdict and permanent injunction), appeal argued Oct. 23, 2024 (2d Cir., pending); Jack Daniel's Properties, Inc. v. VIP Products LLC, 599 U.S. 140 (2023); Abitron Austria GmbH v. Hetronic International, Inc., 600 U.S. 412 (2023); Yuga Labs, Inc. v. Ripps, 144 F.4th 1137 (9th Cir. 2025); MSCHF Product Studio, Inc. v. Vans, Inc. (2d Cir. 2024); Rogers v. Grimaldi, 875 F.2d 994 (2d Cir. 1989); Polaroid Corp. v. Polarad Electronics Corp., 287 F.2d 492 (2d Cir. 1961); Rosetta Stone Ltd. v. Google, Inc., 676 F.3d 144 (4th Cir. 2012); Louis Vuitton Malletier, S.A. v. My Other Bag, Inc., 156 F. Supp. 3d 425 (S.D.N.Y. 2016); Coach Services, Inc. v. Triumph Learning LLC, 668 F.3d 1356 (Fed. Cir. 2012); National Business Forms & Printing, Inc. v. Ford Motor Co., 671 F.3d 526 (5th Cir. 2012); "G.PATTON" decision, Hangzhou Intermediate People's Court (2024).

Statutes: Lanham Act §§ 32, 43(a), 43(c), and definitions — 15 U.S.C. §§ 1114, 1125(a), 1125(c), 1127; Trademark Dilution Revision Act of 2006.

Agency and international materials: U.S. Patent & Trademark Office and U.S. Copyright Office, Non-Fungible Tokens and Intellectual Property: Report to Congress (Mar. 2023); WIPO, Nice Classification (12th ed. 2023; 13th ed. 2024); EUIPO Guidelines for Examination (virtual goods / NFT classification); Korean Intellectual Property Office, Guidelines for Examination of Virtual Goods (July 14, 2022); European Parliament Resolution on the Policy Implications of the Development of Virtual Worlds (Oct. 17, 2024).

Note on currency: Metaverse trademark law is moving quickly. The Second Circuit's MetaBirkins decision, post-Jack Daniel's circuit law, and office classification practice all continue to develop; confirm current authority before relying on any summary here.