Editor’s Note — DCC.
This is DCC’s summary of an opinion piece, “Some Reflections on Data Rights-Confirmation by Registration” (关于数据登记确权的几点反思), by Li Yang (李扬), professor at the China University of Political Science and Law (中国政法大学) School of Civil, Commercial and Economic Law and a vice president of the China Intellectual Property Law Society, published on his personal 李扬知产 account on 24 June 2026. It is a scholar’s argument, not a rule — DCC treats it as a one-off summary rather than a translation, and the framing for overseas counsel is ours.
The timing is what makes it worth reading now. It lands the same week DCC published the Beijing Internet Court’s first application of the AUCL data clause (China’s first AUCL Article 13 ruling), and it is, in effect, the academic case for why that route — not data ownership or a data-IP certificate — is the right one. Read it against DCC’s coverage of the data-IP registration regime: the Datatang v. Yinmu certificate case, what a registration certificate actually confirms, and the registration review guide.
The argument in one line
China keeps trying to “confirm rights” in data (数据确权) and to build data-IP registration on top of that idea — but data has no determinate object, no settled rights content, and no fixed boundaries, and its value lives in constant change, so it cannot satisfy what a confirmation / registration regime structurally requires. The honest path, Li Yang argues, is to drop the ownership frame and protect commercial-data interests under the Anti-Unfair Competition Law.
What “确权” actually means in Chinese law
Li Yang starts by tightening the term. In Chinese IP usage, 确权 (“rights-confirmation”) is not ordinary publicity-effect registration. It is an administrative or judicial act that authoritatively confirms whether an already-existing right is valid — the patent- and trademark-invalidation/confirmation systems are the model. That presupposes three things:
- A determinate object (客体). A patent maps to a specific invention, a trademark to a specific sign, a copyright to a fixed expression — each identifiable and relatively stable.
- Defined rights content (权利内容). The law, not the registration, says what the holder gets (exclusive exploitation, reproduction, etc.).
- Clear boundaries (权利边界). Claims, the registered mark + specified goods, the fixed expression — these draw the line between the right and the public domain. Confirming those boundaries is the whole function.
Confirmation is “determinate object → determinate content → determinate boundary.” It is authoritative recognition of an existing right-state, not the first-time grant of an uncertain interest. Take away that base and “confirmation” has nothing to act on.
Why data fails at the object
- Data is not a legal object by default. No current Chinese statute grants data a unified ownership or analogous right. Civil Code Article 127 is only a referral clause (“where the law provides for the protection of data and online virtual property, those provisions apply”) — it does not say what kind of object data is, what right attaches, or where the boundary runs. So “data IP” is a policy / academic concept, not a strict legal one.
- Data has inherently many producers. Users upload content, merchants supply business information, the platform processes it, algorithms generate labels, third parties add data. You cannot pick a single rights-holder the way you name a patent’s inventor. Registration can show “who submitted the data” — not “who owns it.”
The killer point: a dynamic object can’t be registered
This is the part Li Yang says the debate has underweighted, and it is the strongest move in the piece.
Every registration regime — real estate, IP, equity — presupposes a persistently stable, identifiable object: a registered building isn’t changed by being lived in, a patent isn’t changed by being practiced, a mark isn’t changed by sales. Publicity and reliance work because the object stays put and stays traceable.
Data is the opposite. Its value comes not from any one historical snapshot but from continuous updating — today a new comment, tomorrow a deleted violation, the day after a new merchant. So:
The data at the moment of registration is not the data in reality, and the data in reality is no longer the data that was registered.
Registration demands uniqueness + stability + persistent identifiability; dynamic data clears only the first. When the regime’s assumed object-form collides with how the thing actually behaves, no amount of clever institutional design rescues it.
Why blockchain/hash/timestamps don’t fix it
The standard rejoinder is technical — blockchain存证, hash fixing, timestamps. Li Yang says these miss the level of the problem. A hash proves a given dataset existed at a point in time; it cannot prove the data persists, still less that later, updated data is the same registered object. Blockchain fixes a historical snapshot, not the living data stream.
Technology can prove the past; it cannot lock down the present.
Treating tech as the master key conflates “proof of existence” with “persistent object identification” — two different problems.
What data registration can actually do
Of the four functions usually claimed for data registration, only one survives intact once you take the dynamic-object point seriously:
| Claimed function | Li Yang’s verdict |
|---|---|
| Evidentiary (存证) | Holds up — proves a party held a particular data-form at a point in time. |
| Publicity (公示) | Sharply limited — the object keeps changing, so the register and the real data are substantially decoupled; far weaker than patent/trademark publicity. |
| Transaction support (交易) | Badly mismatched — markets buy the future data stream; the register captures a past snapshot. |
| Judicial proof (司法证明) | Inherently limited — a certificate proves the state at registration, not the state when the dispute arose. |
So registration’s honest value is evidentiary and (weak) publicity/transaction-support — not rights-confirmation.
The pivot: from confirming rights to protecting competition interests
Li Yang names the underlying habit — a “confirmation cult” (确权崇拜): the reflex that every new interest must be “confirmed” to be protected, imported from IP thinking. But IP protects relatively static intellectual outputs; data protection faces a continuously flowing stream. Same logic, wrong object.
His positive proposal tracks what courts already do: data disputes — scraping, crawling — are resolved overwhelmingly through the Anti-Unfair Competition Law, where the court protects not data as an absolute-right object, but the operator’s data competition interest built on substantial investment. That model needs no “data ownership,” no fixed boundaries, and no solution to the dynamic-object problem — which is exactly why it fits the digital economy. He frames the needed shift in three moves:
- from “data rights-confirmation” → “data interest-protection”;
- from static-rights thinking → dynamic-competition thinking;
- from rights-centrism → order-centrism.
And he points to where the statute already says so: Article 13 of the Anti-Unfair Competition Law — the data clause — which, he insists, must not be ignored. The realistic frontier is not an abstract data ownership system but working out the elements of AUCL-protected commercial data and the catalogue of unfair-competition conduct against it.
Why this matters for overseas counsel
- It’s the doctrine under the case. This is the scholarly version of what the Beijing Internet Court just did in China’s first AUCL Article 13 ruling: protect the investment-backed data competition interest without deciding ownership. Li Yang is arguing that route is not a stopgap — it is the correct frame, and the rights-confirmation alternative is built on sand.
- Calibrate what a data-IP certificate is worth. It cuts directly against treating a data-IP registration certificate as title. As DCC has covered, a certificate is evidence of lawful sourcing and investment — useful in an AUCL claim (see Datatang v. Yinmu and what registration actually confirms) — but on Li Yang’s analysis it is not, and cannot be, proof of an exclusive property right in a living dataset.
- Protection is conduct-based, not boundary-based. If the enforceable substance lives in competition law, then what you can stop is improper conduct (circumventing controls, breaching terms, free-riding on another’s investment) — not unauthorized use of “your” data in the abstract. This is the same lesson as the “right to hold data” debate and the three-rights framework: in Chinese data deals, value and protection sit in use/operation + contract + competition law, not in “who owns the data.”
- A health warning on “数据确权.” Expect the term to keep appearing in pilots and local rules; read it, per Li Yang, as registration with evidentiary value — not the conferral of a property right. Don’t advise a client that a registration “confirms ownership” of a dataset.
Source: 李扬, “关于数据登记确权的几点反思”, 李扬知产 (Li Yang IP) WeChat Official Account, 24 June 2026 — original. DCC’s summary and analysis of the author’s argument; not a verbatim translation, and not legal advice.