Your Token Is Not Your JPEG — And That Distinction Is the Entire Point
On the legal void between recognising digital property and actually protecting it
There is a particular species of legal confusion that proliferates wherever technology moves faster than doctrine. It has a recognisable shape. Someone acquires a new kind of asset. Someone else interferes with it. The injured party reaches for a remedy. The law stares blankly. Not because it lacks sympathy, but because every available cause of action was designed for a world that no longer exists in the form the legislature imagined.
We are in that moment now with digital assets, and the confusion is worse than it needs to be — because the people writing about it cannot distinguish three objects that must never be collapsed into one.
The Three Objects
Every dispute involving a non-fungible token involves three things. Not one. Not two. Three. They exist in different technical planes, they are governed by different legal regimes, and the entire legal analysis depends on keeping them separate at every stage.
The work. This is the image, the video, the audio file, the piece of digital media to which the token refers. It lives off-chain — on a server, on IPFS, on some other storage layer. It is infinitely reproducible. Copying it costs nothing and affects the token not at all. Its legal protection lies in copyright. Always has. Always will.
The token. This is the on-chain ledger entry. A unique identifier, a contract address, an owner address. It cannot be duplicated on-chain: the protocol enforces non-fungibility by design. Control is exercised through the private key. The key-holder can transfer it, burn it, or hold it — and no intermediary’s consent is required. The key-holder’s position is exclusive, rivalrous, and self-executing.
The conduct. This is the act alleged to constitute wrongful interference. It may be directed at the work (copying it), at the token (transferring, freezing, or destroying it), or at neither (merely viewing publicly displayed content). The legal analysis depends entirely on which layer the conduct targets.
Collapse these layers and you produce the confusion that saturates this field. Copyright injuries are dressed up as property claims. Platform misconduct against the work is mistaken for interference with the token. Economic harm to the market environment is treated as dispossession. None of this is correct. The distinction is not a matter of degree. It is categorical.
The relationship between the work and the token is referential, not constitutive. Possessing the token does not confer copyright in the work. Possessing the copyright does not confer any rights over the token. The Hangzhou Internet Court understood this in the Fat Tiger case — the first Chinese court ruling on an NFT — when it characterised the token as having property interests (财产权益) distinct from the copyright dimension of the associated artwork. It then resolved the dispute on copyright grounds and left the property question untouched.
That untouched property question is where the real legal problem begins.
The Gap
Both China and Singapore have recognised that digital assets possess something like a proprietary dimension. Chinese courts have treated crypto assets as having property attributes. The Singaporean High Court in CLM v CLN held that cryptocurrencies satisfy the four Ainsworth criteria for property. The Court of Appeal in Quoine observed — without deciding — that there is ‘much to commend’ the assimilation of cryptocurrencies into general property concepts.
Recognition achieved. Problem solved?
No. The problem starts here.
Recognition of a proprietary dimension does not determine which tortious remedies attach. Both systems have said: this looks like property. Neither system has said: and here is what happens when someone interferes with it.
That is not a theoretical gap. A Chinese holder of a digital collectible NFT whose token is transferred without authorisation by a platform or hacker faces a concrete remedial problem. If the token qualifies as a ‘thing’ (物) under Articles 114–115 of the Civil Code, the holder can seek return of the token without proving fault, elimination of interference, restoration, and damages. If it does not — because Article 127’s delegation to future legislation has never been fulfilled — the holder is limited to copyright, contract, or general fault-based tort claims. The practical difference: the property-protection remedies do not require fault for possessory relief. The general tort claim requires fault for everything, including return.
In Singapore, the holder faces a different version of the same problem. The tort of conversion protects possessory interests in personal property — but it has historically required tangible subject matter. An NFT is not tangible. No Singaporean court has extended conversion to digital tokens. Until one does, the holder cannot bring a conversion claim at all.
Two systems. Two different doctrinal architectures. The same practical result: holders of digital assets lack a settled tortious remedy for possessory interference.
Why the Gap Exists
The institutional reasons differ, and they are worth stating precisely because they tell you something about how legal systems actually work — as opposed to how law professors wish they worked.
In China, the gap is legislative. The Civil Code delegates the protection of ‘network virtual property’ (网络虚拟财产) to future legislation via Article 127. Courts operating under the numerus clausus principle — the categories of property rights are fixed by law — are institutionally reluctant to create new property-right categories that the National People’s Congress has not authorised. So they do what Chinese courts characteristically do when the legislature has deferred: they recognise property attributes in individual cases without formally classifying the asset as a ‘thing’ under the property-rights framework. Pragmatic. Incremental. Unresolved.
In Singapore, the gap is litigation-driven. Common-law courts extend torts incrementally, and the extension of conversion to a new class of subject matter requires a case that squarely presents the question. No such case has arisen. The claimant in CLM sought a proprietary injunction, not conversion damages. Until someone brings a conversion claim against a defendant who dealt with a specific digital token, the courts will not address the question.
The Chinese gap is about legislative authority. The Singaporean gap is about litigation opportunity. Both produce the same void.
What Is Not a Property Case
Here is the part most people get wrong, and it is the part that matters most.
The overwhelming majority of NFT disputes involving screen scraping, content copying, and marketplace aggregation are not possessory-interference cases. They are copyright cases, computer-misuse cases, unfair-competition cases. They belong in those doctrines and should stay there.
The temptation to reach for property-interference remedies when copyright seems inadequate or when the economics of digital scarcity have been undermined must be resisted. Property-interference liability protects a possessory interest in a specific asset. It does not protect the economic context in which that asset has value.
Consider the families of conduct:
Family 1: Passive copying. Someone accesses publicly displayed content associated with NFTs and copies images or metadata. The token remains on the ledger, under the same key-control, entirely unaffected. This is a copyright claim. Neither system’s property-interference liability is engaged because there has been no dealing with the token.
Family 2: Access-control circumvention. Someone bypasses platform controls — credential stuffing, API abuse, session hijacking — to extract off-chain content. The additional wrongfulness of the means does not engage property-interference liability if the token itself remains under the holder’s control. Wrongful means do not create possessory wrongs.
Family 3: Token-level interference. Someone transfers the token to another wallet without authorisation. A platform freezes its transferability. A custodial provider refuses to process instructions. The token is burned to an inaccessible address. This is the only category in which property-interference liability is engaged. The conduct must be directed at the token, not at the associated content.
The indirect-interference argument — that systematic scraping destroys the token’s commercial environment and therefore harms the token itself — fails. The tokens remain on the ledger, under the same key-control, technically unimpaired. Reducing the market value of a token by producing competing content is market competition, not property interference. The tort protects the thing, not the surrounding market conditions in which the thing happens to be valuable.
This negative conclusion is more useful than any affirmative extension. It tells courts and practitioners where not to go. It keeps copyright in copyright, computer-misuse in computer-misuse, and unfair competition in unfair competition. The inflation of property remedies to cover every form of digital economic harm is precisely the doctrinal mistake the three-layer separation is designed to prevent.
Why Property-Interference Liability and Not Something Else
If the test excludes most disputes, why does the residual category matter?
Because adjacent doctrines fail precisely where token-level interference is most serious.
Contract requires privity. A hacker who transfers a token without authorisation has no contractual relationship with the holder. Unjust enrichment is restitutionary — it reverses gains but does not vindicate possessory control over a specific asset. Unfair-competition law protects market order, not exclusive control over an identified thing, and typically requires a competitive relationship that does not exist between a holder and a thief. Equitable tracing follows proceeds but does not provide a direct claim for return of the token itself. Criminal prosecution punishes but does not restore.
In Chinese law, only the property-protection remedies of Articles 235–237 permit return and elimination of interference without proof of fault. In Singaporean law, only conversion provides a cause of action requiring neither contractual privity, nor competitive relationship, nor unjust enrichment — only that the defendant dealt with the claimant’s property seriously enough to justify forced-sale damages.
The residual category is small. The remedial gap within it is not.
Two Systems, Two Architectures
If property-interference liability does attach, the remedial consequences differ sharply between the two systems — and that divergence is itself illuminating.
Chinese law provides a modular remedial menu. Article 235 permits the return of the original thing: if the wrongfully transferred token can be traced to a specific wallet, the court can order its return. Article 236 permits the elimination of obstruction: if a platform has frozen transferability, the court can order the freeze lifted. Article 237 permits restoration. Article 238 adds damages where the infringement causes loss and the actor is at fault. These remedies can be applied individually or in combination. The claimant does not need to prove fault to obtain possessory relief — only for compensatory damages. A Chinese court could order the return of a wrongfully transferred token even if the court could not award damages because fault was not established. No equivalent bifurcation exists in the common law.
Singaporean conversion operates on an all-or-nothing logic. The standard measure of damages is the fair market value of the chattel at the time and place of conversion. Title to the converted chattel passes to the defendant upon payment. In the digital-asset context, this means the defendant — possibly a hacker with no legitimate use for the token — acquires the token upon paying its value. That is commercially incoherent. Conversion’s strict liability eliminates the need to prove fault, which is an advantage where the defendant’s identity is obscured by anonymised blockchain transactions. But the rigid forced-sale measure creates valuation problems in volatile NFT markets that Chinese law’s loss-based assessment avoids.
The structural lesson: the gap between recognition and remedy has different practical consequences in each system. Closing it would produce different remedial architectures. That is why the comparative analysis matters — not because the systems converge, but because each system’s response reveals something about the structure of the problem that the other system’s response conceals.
The Formalist Objection — Stated Honestly
The formal obstacles are real and must not be trivialised.
In China, numerus clausus means the categories of property rights are fixed by law. Even a fully functional digital equivalent of a chattel cannot become a property-right object until the National People’s Congress says so. Article 127’s delegation is precisely a statement that it has not yet said so.
In Singapore, tangibility is part of conversion’s institutional identity. The tort is, by nature and history, a tort about physical things. Extending it to non-physical things changes what it is, not merely what it covers.
These are not quaint relics awaiting enlightened modernisation. They are constitutive limits. If a court accepts them as such, it will decline extension. That court will then face the practical question: what remedial architecture applies to digital-asset interference in the absence of property-protection or conversion liability? Neither system has answered that question either.
The functional response is that each formal barrier has a rationale, and the rationale can be assessed. In China, the purpose of numerus clausus is to prevent the uncontrolled creation of proprietary burdens that bind third parties. A key-controlled digital token, held by a single address on a public ledger, creates no greater burden on third parties than a chattel held in a locked room. If the constraint is about third-party burden, the constraint is satisfied. In Singapore, the purpose of tangibility was to confine conversion to objects susceptible to possessory control. A key-controlled digital token is susceptible to possessory control — more so than many tangible chattels, since the holder’s control is technically absolute rather than dependent on physical custody.
Whether courts accept this reasoning is an open question. That it is a serious question — rather than a rhetorical one — is the point.
The Test
Five steps. Applicable in both jurisdictions. Designed to force the decision-maker to ask the right questions in the right order.
Step 1. Identify the layer. Is the alleged wrong directed at the work, the token, or neither? If at the work, route to copyright. Proceed only if directed at the token.
Step 2. Identify the object. Does the token possess the characteristics of a ‘thing’ under the Chinese Civil Code? Does it satisfy the Ainsworth criteria under Singaporean law? If not, no property-interference liability.
Step 3. Assess whether exclusive control was impaired. Did the defendant’s conduct impair the holder’s ability to transfer, use, and exclude? Copying off-chain content does not impair key-holder powers regardless of economic harm.
Step 4. Assess seriousness. In China: fault for damages, but no fault required for possessory relief. In Singapore: permanent deprivation meets the conversion threshold; temporary interference does not.
Step 5. Route. If Steps 1–4 are satisfied, property-interference liability or conversion. If any step fails, redirect to copyright, computer-misuse, contract, unjust enrichment, or unfair competition.
The test excludes claims that do not involve interference with exclusive control, regardless of the economic harm caused. That exclusion is the test’s most important function.
What Follows
The question extends beyond NFTs. Any digital asset exhibiting exclusive cryptographic control, rivalrousness, and identifiability — tokenised securities, decentralised-finance positions, digital identity credentials, on-chain governance rights — presents the same doctrinal question.
Chinese law and Singaporean law face it from different starting points. Both have recognised the property dimension. Neither has specified the tortious consequence.
The implication is straightforward. Legal systems designed for physical objects must decide what follows when economically significant assets are exclusively controlled through technical means but lack physical form. The answer is not that everything becomes property. Most of it does not. The answer is that private law must identify, with precision, the narrow class of conduct that actually interferes with token-level control — and provide a remedy for it.
The three-layer distinction tells you where to look. The typology tells you where not to look. The test routes the claim. Everything else is noise.
The future meaning of possession in private law depends on whether legal systems can draw that line without either refusing to protect digital assets at all or dissolving the boundary between possessory-interference liability and everything else. Both errors are available. Both are already being made.
The first error — refusal — leaves holders of valuable digital assets without a tortious remedy when their tokens are stolen, frozen, or destroyed. It tells them to go to contract when there is no contract, to unjust enrichment when the problem is not enrichment but dispossession, to criminal law when what they need is return. That is not legal conservatism. It is legal abdication.
The second error — dissolution — turns every digital economic grievance into a property claim. It swallows copyright, unfair competition, and market-regulation disputes into a tort that was never designed to carry them. It produces the inflation of possessory remedies that the three-layer separation is built to prevent. That is not legal modernisation. It is doctrinal vandalism.
The discipline required is to avoid both. To protect the token without pretending the token is the work. To recognise possessory control without abolishing the limits that make possessory remedies coherent. To say clearly: this narrow class of conduct — unauthorised transfer, destruction, impairment of key-holder powers — engages property-interference liability. Everything else does not.
That is not a compromise between conservatism and innovation. It is the application of private-law logic to a new category of object. The logic is old. The object is new. The analysis, if done properly, is straightforward.
Most people writing in this area are not doing it properly. They are either terrified of the technology or intoxicated by it. Both states produce bad law. What is needed is neither terror nor intoxication but the ordinary, rigorous, unglamorous work of doctrinal classification.
Identify the layer. Identify the object. Assess interference. Route the claim. Done.