Transferring the intangible: applying Rome I to the sale of non-fungible tokens

NFTs are widely described as electronic records ‘capable of being subject to control’ that fall within the broader understanding of digital assets as items represented digitally or electronically (Ripley, 2024). Technologically, NFTs are unique digital records minted through smart contracts, most commonly on the Ethereum blockchain, following token standards such as ERC-721. The token encodes its defining characteristics, such as provenance, uniqueness and owner, ensuring their persistence and immutability within the blockchain infrastructure. In most cases, however, the NFT does not contain the underlying digital file (e.g., artwork or video) but rather points to it via metadata. This separation between token and asset highlights the fragility of the code-content relationship: if the linked metadata becomes inaccessible or corrupt, the NFT may lose its intended functionality or market value (Martínez Luna et al., 2024).

NFTs are stored in digital wallets and controlled by private cryptographic keys, allowing peer-to-peer transfers without reliance on traditional intermediaries (Falokun, 2023). Their decentralised structure challenges traditional localisation rules, as NFTs are not confined to any physical jurisdiction but exist across a distributed ledger—simultaneously ‘everywhere and nowhere’ (Law Commission, 2024). At the same time, the immutability of blockchain records generates a form of digital scarcity, granting NFTs the attributes of rarity and exclusivity in ways previously unavailable to purely digital items (Martínez Luna et al., 2024).

The uses of NFTs now extend well beyond digital art. They support gaming economies, metaverse real estate and programmable financial instruments. They operate as access credentials, digital identifiers or instruments within decentralised finance (DeFi) ecosystems (Ante and Fiedler, 2025; Lehmann, 2024). This functional heterogeneity explains why NFTs resist assimilation into traditional legal categories. However, classification is essential under the Rome I Regulation, which links the applicable law to contract type.

Doctrinal views on the legal character of NFTs diverge. One approach argues that NFTs may be analogised to goods, based on their uniqueness, transferability and blockchain-enforced exclusivity, which collectively replicate aspects of tangible ownership. However, as Ripley notes, blockchain assets are purely digital and lack physical/tangible form (Falokun, 2023; Ripley, 2024). This position aligns with the case law of the CJEU, which has defined ‘sale of goods’ narrowly, requiring the transfer of ownership of tangible movables (CJEU, Car Trim, C-381/08). Although this interpretation had arisen in the context of jurisdiction, this definition should be considered relevant for analogous provisions in Rome I, given the tendency to interpret harmonised concepts consistently across the EU’s private international law instruments.

An alternative and more widely supported view treats NFTs as intangible records of rights or data entries. Ripley emphasises that blockchain-based rights often blur the line between proprietary and obligatory entitlements (Ripley, 2024). Comparative analysis confirms that many legal systems hesitate to recognise NFTs as property in the classical sense, since they are neither choses in possession nor conventional choses in action (Lee, 2024).

Some scholars propose hybrid or functional approaches, acknowledging that NFTs blur traditional categories. The concept of digital ‘ownership’ is fragmented: while a user may control the token, the associated file, platform terms and copyright entitlements often remain external. Lehmann (2024) argues that digital scarcity and new ways of value representation challenge existing categories and call for a functional analysis of the contractual or transactional relationship underlying the NFT. Others have similarly pointed out that digital assets are ‘omniterritorial’—resisting localisation within a single jurisdiction due to their distributed nature (Garrido, 2023; Ripley, 2024). Adding to these complexities, some authors question whether frameworks like Rome I can adequately accommodate NFTs at all, given the structural mismatch between decentralised, borderless systems and territorially grounded conflict rules (Falokun, 2023).

This doctrinal fragmentation has direct consequences for Rome I. The Regulation applies only to contractual obligations, and its default rules depend on how a given contract is classified. If the sale of an NFT could be characterised as a contract for the sale of goods, Article 4(1)(a) would apply. If instead it is regarded as a service contract, Article 4(1)(b) becomes relevant. Where neither classification is appropriate, the residual rule in Article 4(2) points to the law of the party effecting the characteristic performance (since other categories under Article 4(1) are inapplicable to NFT sale agreements per their scope). The choice of category thus directly shapes the applicable law.

CJEU jurisprudence narrows these options. Since ‘goods’ are limited to tangible movables, the more plausible classification under Rome I is that of services. While the term ‘services’ is not expressly defined in Rome I, interpretative guidance can be drawn from the CJEU’s definition in Falco (CJEU, C-533/07), delivered under the Brussels I Regulation. In that case, ‘services’ were described as a positive activity performed for remuneration. Although this definition arises in the context of jurisdiction, it has been widely adopted in legal scholarship and commentary on Rome I, due to the consistent interpretative approach generally applied across EU private international law instruments. Transposing this reasoning, the transfer of an NFT may constitute such a positive activity—particularly where execution is initiated by the seller or mediated by a platform. Indeed, current academic consensus leans towards treating NFT sales as service contracts under Article 4(1)(b), with Article 4(2) serving as a fallback where pseudonymity or automation obscures the parties. In sum, while Rome I can be applied to NFT sale contracts, this application depends on analogies that expose doctrinal fragilities. The definitional ambiguity of NFTs strains traditional classification under the Regulation. These foundational uncertainties underscore the need for careful, context-sensitive interpretation when extending Rome I to digital assets.

The Rome I Regulation governs ‘contractual obligations in civil and commercial matters’ that involve a conflict of laws (Article 1(1)). In order to determine whether a claim falls within this scope, the CJEU has consistently held that two conditions must be satisfied: first, the obligor must have freely assumed the obligation vis-à-vis the obligee (CJEU, Handte, C-26/91); and second, the assessment must focus on the elements that characterise the nature of the legal relationship between the parties and the subject matter of the dispute (Weller, 2015). NFT sale agreements, although technologically novel, are contracts in which parties agree to transfer a tokenised digital asset in exchange for consideration. Their contractual dimension therefore falls squarely within the substantive scope of Rome I.

The interpretative structure of Rome I is well established. The principle of party autonomy, enshrined in Article 3, is the cornerstone of the Regulation: the parties are free to choose the law applicable to their contract. If no valid choice is made, the applicable law must be determined according to the typology set out in Article 4. In line with the European tradition, Rome I thus accords a central role to the autonomy of the contracting parties (Magnus and Mankowski, 2017). The threshold question in the context of NFT sales is whether, and how, such autonomy can be exercised in decentralised, pseudonymous and platform-mediated environments.

Article 3(1) permits parties to choose the applicable law ‘expressly or clearly demonstrated by the terms of the contract or the circumstances of the case.’ The Giuliano-Lagarde Report underlines that this choice must be ‘real’, and not presumed lightly. In conventional practice, party autonomy is most commonly exercised by including a governing-law clause within the contract or a related document (Giuliano and Lagarde, 1980). Implied choices, by contrast, must be inferred from the terms or circumstances in a manner that clearly demonstrates the parties’ intention (Magnus and Mankowski, 2017).

This framework extends to smart contracts, provided they qualify as contracts in the legal sense. Harrak (2023) observes that smart contracts, including those widely used for NFT sales, ‘should be regarded as legal contracts and therefore legally binding if there is an offer and a corresponding acceptance in the legal sense’. If characterised as contracts, they fall within the scope of Article 3 and may incorporate a valid choice of law. However, as Kleczewski (2023) emphasises, the pseudonymous nature of blockchain transactions complicates the identification of the parties and the ascertainment of contractual intent—factors which make it especially difficult to infer a ‘clearly demonstrated’ choice. The demanding threshold set by Article 3(1) is thus particularly challenging to satisfy in such environments.

NFT sales are frequently conducted on online marketplaces, including platforms such as OpenSea, Rarible, Crypto.com and Binance (Tan, 2024). These platforms typically impose Terms of Service (ToS), which often contain governing-law clauses. For example, OpenSea’s Terms of Service (June 2025) provide: ‘These Terms and your access to and use of the Service shall be governed by and construed and enforced in accordance with the laws of the State of New York’ (OpenSea, 2025). The doctrinal question is whether acceptance of such ToS—typically via clickwrap mechanisms—can qualify as a valid choice of law under Article 3.

Some commentators confirm that standard form terms may be effective if assent is clearly established (Magnus and Mankowski, 2017). This view finds further support in soft-law initiatives: the UNIDROIT Principles on Digital Assets provide that a choice of law expressed in the asset itself or in the system where it is recorded may be effective (UNIDROIT Principles 2023, Principle 5(1)(a)-(b)). Thus, ToS could in principle constitute a valid exercise of party autonomy under Rome I.

Nonetheless, both doctrinal authorities and CJEU case law emphasise the limits of such reasoning. Passive acquiescence does not suffice: a governing-law clause must reflect genuine agreement. The CJEU has indicated that mere inclusion of such clauses in standard terms—particularly in consumer contracts—may not withstand scrutiny where there is no active, informed consent (CJEU, Verein für Konsumenteninformation, C-191/15). Magnus and Mankowski (2017) similarly caution that ‘mere submission is not positive assent’. In decentralised NFT marketplaces, where assent is often reduced to pseudonymous clicks, the legal efficacy of ToS-based choices of law remains uncertain.

As said, additional complexity arises when NFTs are purchased by consumers. Article 6 of Rome I mandates that any choice of law must not deprive consumers of protections afforded by the law of their habitual residence. In Verein für Konsumenteninformation, the CJEU held that a governing-law clause in standard terms may be found unfair if it creates the false impression that only foreign law governs the contract, misleading consumers about their rights under Article 6(2) (CJEU, Verein für Konsumenteninformation, C-191/15).

Beyond platform terms, a further question is whether a choice of law may be embedded in the architecture of a smart contract itself. Rühl (2021) argues that algorithmic execution—‘if this, then that’—cannot itself constitute a juridical act of choosing applicable law. Instead, the choice must be expressly declared, either within the legal contract executed by the smart contract, or in a distinct instrument such as a Ricardian contract (a document recorded in both human-readable and machine-executable format, linked to the relevant digital asset). While Article 3(1) allows for implied choices, these must be ‘clearly demonstrated’ from the terms or circumstances of the case. This presupposes evidence of the parties’ actual intention— an element often absent in blockchain environments, where many participants are unaware that such a choice may even be made. While the European Law Institute (ELI) Principles on Blockchain Technology and Smart Contracts focus primarily on consumer protection and contract law, they nonetheless reinforce a key insight relevant to Rome I: blockchain transactions do not exist outside legal frameworks. The ELI Principles confirm that ordinary conflict-of-laws rules remain applicable, including the requirement that any choice of law be clearly expressed or demonstrable (European Law Institute, 2023, Principle 4), suggesting that similar conclusions could also be drawn regarding the NFTs.

The foregoing reveals the fragility of party autonomy in NFT sales. While Article 3 of Rome I accommodates the possibility of choosing applicable law, the pseudonymous and decentralised character of many transactions undermines the valid operation of this principle. Platform terms may satisfy the requirements of express or implied choice where properly incorporated, but concerns remain about effective consent, especially where clickwrap mechanisms obscure legal consequences. Smart contracts, though efficient for performance, rarely substitute for a deliberate act of governing-law selection.

Where no valid choice is established, the default connecting factors under Article 4 govern. In this context, the consumer protection framework of Article 6 retains particular importance, given the prevalence of NFT markets directed at non-professional users. Where the buyer qualifies as a consumer, the parties’ choice of law may not deprive the consumer of the protection afforded by the law of their habitual residence (Article 6(2), Rome I). Given that NFT platforms often operate globally and use standardised digital terms, this rule acquires practical relevance. Governing-law clauses buried in clickwrap ToS may be unenforceable if they mislead or disadvantage the consumer. In addition, overriding mandatory provisions under Article 9 Rome I may intervene—particularly where domestic laws governing anti-money laundering, taxation or crypto-trading regimes impose non-derogable requirements that affect the enforceability of NFT sale contracts. Taking this all into consideration, it is no surprise that some scholars have been prompted to advocate for a more fundamental shift: the development of digital-specific rules of applicable law, better tailored to blockchain environments (‘Lex Cryptographia’ and ‘Lex Metaversi’), rather than bending Rome I’s 20th-century logic to accommodate 21st-century technologies (López Rodríguez, 2024).

Where party autonomy under Article 3 Rome I is unavailable or fails to produce a valid choice, the determination of applicable law falls to the default framework under Article 4. That framework operates as a cascade: first, presumptive rules by contract type (Article 4(1)); then the fallback to the law of the characteristic performer (Article 4(2)); the exception for manifestly closer connection (Article 4(3)); and, in the most complex cases, an open-ended closest connection test (Article 4(4)). This structured hierarchy aims to balance legal certainty with contextual flexibility. Whether this architecture can accommodate NFT sale contracts is the central question of this chapter.

The starting point under Article 4(1) is classification. Can an NFT sale be assimilated to a contract for the sale of goods (Article 4(1)(a)) or the provision of services (Article 4(1)(b))? Other categories under Article 4(1)(c)-(h) are inapplicable to standardised NFT sale agreements. As discussed previously, ‘goods’ under Rome I is interpreted autonomously and narrowly. Since NFTs are intangible, ledger-based data structures, the sale of such tokens cannot be brought within Article 4(1)(a). As Rühl (2021) and Mariański (2024) emphasise, market terminology describing NFT transfers as ‘sales’ does not override the Regulation’s strict legal classifications.

The prevailing doctrinal position favours classifying NFT transactions as contracts for services under Article 4(1) (b). This rests on a functional understanding of the seller’s obligation: the initiation and execution of an on-chain transfer, typically in exchange for payment, constitutes a positive activity performed for remuneration. Wendehorst (2024) argues that such classification best aligns with the Regulation’s logic when tangible subject matter is lacking. Mariański (2024) concurs, noting that a consistent application of the service model offers a pragmatic solution that supports legal predictability.

Still, this analogy is not without strain. Certain NFT sales execute automatically through smart contract code, without real-time human intervention. In these contexts, the concept of ‘performance’ becomes abstract: if no party actively executes an obligation, the foundational premise of a service relationship may erode (Ripley, 2024). Thus, while the service classification remains doctrinally viable, it is an approximation rather than a perfect doctrinal fit.

Comparative frameworks have begun to respond to these limitations by proposing alternative models. The UNIDROIT Principles on Digital Assets (2023) adopt a system-centric approach: the applicable law may be that expressly declared within the digital asset itself (e.g. via metadata) or, failing that, within the system in which the asset is recorded. Only where no such designation exists do residual rules apply, allowing the forum to rely on its own law or general principles of private international law (Principle 5). This demonstrates a possible movement away from rigid goods-versus-services dichotomies and towards digital-specific connecting factors.

Assuming the classification as a service contract fails as well, the next step is to determine the characteristic performance under Article 4(2). In conventional terms, the seller’s obligation to deliver constitutes the characteristic performance in a sale (Opinion of Advocate General, Falco, C-533/07). By analogy, in NFT sales, this would be the seller’s action of transferring the token to the buyer. However, two practical complications arise. First, pseudonymity obscures the identity of the performer. Wallet addresses lack legal personality, and there may be no verifiable real-world identity linked to the transaction. Second, sales are often mediated by platforms such as OpenSea or Rarible, which structure the transaction, impose terms and in some cases even handle custody or escrow mechanisms. Some commentators argue that such platforms act as the characteristic performer, particularly where they play a determinative role in facilitating the transfer (López Rodríguez, 2024).

This raises the question of whether Rome I’s structure can accommodate a shift in focus from the contractual party to the intermediary. While treating the platform as the characteristic performer may offer a functional anchor, it undermines the principle that the connecting factor should relate to the contracting parties themselves. Moreover, this model does not apply in peer-to-peer transactions, where no intermediary is involved. The result is a fragmented application of Article 4(2) that fails to generate consistent outcomes.

Further complexity emerges from blockchain’s delocalised architecture. Although ‘delivery’ has no autonomous meaning under Rome I, it is relevant insofar as it informs the identification of the characteristic performance. In blockchain transactions, the ‘delivery’ of an NFT is completed by updating the ledger through a consensus mechanism. Yet this process has no physical location. Ripley describes it as ‘executed and recorded on the ledger’, without an analogue to localisable performance (Ripley, 2024). Rühl (2021) likewise warns that this absence of situs destabilises the territorial assumptions embedded in Rome I. If the place of performance cannot be established, the Regulation’s spatial logic begins to unravel.

Where Articles 4(1)(b) and 4(2) fail to produce a workable connecting factor—either because the performer is unidentifiable or the legal characterisation is uncertain— parties must turn to Articles 4(3) and 4(4) Rome I. Article 4(3) permits a departure from the default rule if the contract is manifestly more closely connected with another country. This allows courts to override artificial results produced by misclassification or pseudonymity. Article 4(4), in turn, provides a final recourse where the contract type is not listed in Article 4(1) or determined per Article 4(2), empowering courts to apply the law of the country most closely connected to the contract. While these provisions grant flexibility, they do so at the expense of predictability. In pseudonymous or cross-jurisdictional transactions, it may be impossible to determine what legal system is ‘most closely connected’ to the contract (Ripley, 2024). UNIDROIT’s system-centric approach model, in the author’s opinion, aims to offer more structured alternatives, suggesting that Rome I’s fallback clauses may no longer suffice for digital-only transactions.

The operational challenges of applying Rome I become particularly apparent when tested against two common transaction types:

• (a)

  Example A: Platform-mediated sale with governing-law clause. A Latvian buyer purchases an NFT through a Dutch-based NFT marketplace. The platform’s ToS include a Dutch law clause, accepted via clickwrap. After a technical failure in delivery, a dispute arises. Under Article 3 of Rome I, Dutch law applies if the clause is validly incorporated and clearly demonstrated. However, if the buyer is a consumer, Article 6 intervenes: the choice may not deprive them of protections under Latvian law if the platform directs activity to Latvia. If the clause is deemed invalid or unclear, Article 4(1)(b) may apply, identifying the seller’s action as characteristic performance. In practice, this may require identifying the habitual residence of the seller—assuming the pseudonymity barrier can be overcome. Article 9 could also allow mandatory Latvian provisions to apply if they concern NFT regulation.

• (b)

  Example B: Pseudonymous peer-to-peer sale without terms. Two pseudonymous users from two EU countries transact directly via a decentralised protocol. There is no platform, no governing-law clause, and the parties’ identities or locations are unknown. A dispute arises due to metadata failure. Party autonomy under Article 3 cannot operate due to a lack of demonstrated choice. The performer is unidentifiable, and the performance is technically executed by a protocol; hence, Article 4(2) fails. Article 4(3) or (4) may be invoked, but the forum will face serious difficulty in applying a closest connection test to a transaction with no discernible geography or real-world identity. Legal certainty under Rome I, in this context, depends on factors that may not exist.

Although Article 4 governs the designation of applicable law, other provisions of Rome I also deserve mention. As discussed previously—Article 6 ensures that consumers remain protected by the mandatory rules of their habitual residence, even where a choice of law or default rule would otherwise apply. Additionally, Article 9 Rome I permits the application of overriding mandatory provisions of the forum, or of another country with a close connection to the case. Courts applying Rome I may therefore be required to give effect to such laws even where the default or chosen applicable law would otherwise not permit them. Articles 5 (carriage) and 7–8 (insurance and employment) are not directly relevant to NFT sales. Once applicable law is determined, Article 12 defines its substantive scope: it governs interpretation, performance, remedies and the consequences of breach. In the NFT context, this includes assessing whether the transfer was validly executed, whether it can be reversed or rectified in cases of technical error, and what remedies may be available under the applicable contract law.

The analysis above shows that while Rome I can, in principle, be applied to NFT sales, its operation is fragile and doctrinally strained. The prevailing classification of such contracts as services under Article 4(1)(b) offers a presumptive anchor, but pseudonymity makes identifying the characteristic performer and their habitual residence difficult, and in peer-to-peer transactions, often impossible. Attempts to rely on platforms as proxies raise concerns about the proper subject of the connecting factor. The fallback rules of Articles 4(2)–(4) provide some flexibility but at the cost of predictability, particularly in decentralised transactions with no territorial connection. Comparative initiatives illustrate that alternative issuer-centric or system-centric approaches may provide greater certainty, but they also highlight how far Rome I must be stretched to accommodate digital-only transactions. Rome I remains formally applicable, but its capacity to ensure legal certainty in decentralised transactions is increasingly fragile—a tension that calls for critical reassessment considering 21st-century technological realities.

The preceding analysis confirms that while Rome I is formally applicable to cross-border contracts for the sale of NFTs, its underlying architecture is increasingly misaligned with the characteristics of digitally native transactions. The Regulation presupposes a legal environment structured around territorially anchored performance, predefined contractual relationships and identifiable legal subjects. In NFT markets, these premises are no longer reliably present.

NFT markets operate within ecosystems where pseudonymity, automation and functional modularity replace bilateral intention and identifiable subjectivity. The Regulation’s default categories retain coherence in the analogue world but lose doctrinal traction when applied to tokenised, smart-contract-based exchanges. As examined previously, the qualification of an NFT sale as a ‘service’ under Article 4(1)(b) may be doctrinally justifiable, yet remains an approximation. Execution is often automated, irreversible and devoid of continuous human input—challenging the classical understanding of performance as a positive, relational act (Mariański, 2024; Ripley, 2024).

A more fundamental friction arises from the Regulation’s reliance on connecting factors—most notably, habitual residence. Rome I assumes that contracting parties can be located. In peer-to-peer NFT transactions, the parties may be pseudonymous, and no legally relevant location may be ascertainable. As Kleczewski (2023) notes, even where metadata or technical identifiers exist, translating these into enforceable legal presence remains highly uncertain. Without a determinable habitual residence, the law of the ‘characteristic performer’ cannot operate as a predictable anchor.

This fragility affects legal certainty. Rome I aspires to provide foreseeability in cross-border transactions. Yet when the connecting factors yield indeterminate outcomes—whether due to pseudonymity, decentralisation or the automation of obligations—courts may default to the ‘manifestly closer connection’ exception (Article 4(3)) or the open-ended test of closest connection (Article 4(4)). While permissible, these fallback mechanisms risk becoming discretionary rather than jurisprudential, particularly in the absence of concrete territorial or personal links. Legal predictability—central to the legitimacy of private international law—therefore becomes increasingly difficult to secure in such environments.

This does not imply that Rome I produces arbitrary outcomes. Rather, it highlights that the Regulation’s logic—connecting legal obligations to jurisdictional anchors—presumes a transactional architecture that digital asset environments increasingly resist. European private international law is built upon foundational concepts: territoriality, personality and bilateral intention. NFT markets displace these anchors. Transactions are automated, parties are anonymised and platforms may merely facilitate rather than contract. Rome I, when applied to such contexts, begins to stretch beyond its intended doctrinal design.

Comparative instruments illustrate how legal design may begin to respond. The UNIDROIT Principles on Digital Assets and Private Law propose a system-centric approach, allowing applicable law to be embedded in the digital asset or the infrastructure on which it is recorded (UNIDROIT Principle 5). This proposal should not be overstated as immediate alternative to Rome I. It operates in parallel and offers comparative insight rather than doctrinal displacement. Nonetheless, it serves as persuasive indicator that legal certainty in decentralised environments may increasingly depend on frameworks that accommodate their operational logic.

Against this backdrop, there is growing support for a calibrated EU-level response. The challenges posed by NFTs are not isolated—they are emblematic of broader transformations in how digital value is structured and transacted. On-chain transfers are already central to markets in digital art, metaverse property, DeFi and emerging forms of tokenised equity. If left unaddressed, the interpretative strain on existing private international law instruments may deepen.

Rome I remains a fundamentally resilient and adaptable regulation. Yet to preserve its relevance, three reform trajectories merit consideration. First, interpretative clarification by the CJEU could provide doctrinal guidance on how concepts such as ‘habitual residence’ or ‘characteristic performance’ should be understood in blockchain contexts. Second, the European Commission could issue soft-law guidance, such as a communication, clarifying the application of Rome I to digital asset (or NFT specific) transactions. Third, in the longer term, the EU may consider either amending Rome I or developing a bespoke conflict-of-laws instrument tailored to digital assets (or NFTs specifically). These options are not mutually exclusive. A layered approach by combining jurisprudential development, soft-law support and targeted legislative innovation could ensure that European private international law remains responsive, conceptually sound and functionally effective in the digital era.