Filed under academic-commentary
Every brief tagged "academic-commentary".
- § 01 · DATA-PROPERTY-RIGHTS
Data 'Parallel Property Rights' — They Can Confer Status, but Can't Secure Control
Part four — and the synthesis — of Hong Yanqing's (洪延青, 网安寻路人) study notes on China's 'separation of three rights' data-property framework takes up 'parallel property rights' (数据平行财产权): how to allocate rights when the *same* data is held, used, and operated by *multiple* parties at once. Building on Xiong Bingwan and Zhuang Hongshan's 'one-data, multiple-rights' (一数数权) idea — data is non-rivalrous and copyable, so the same right over the same data can sit with several parties without excluding each other — Hong argues parallel property rights are best understood as *default rules* for incomplete-contract, collaborative-production settings: internally, parallel use is presumed; externally, operation is classified by data type (by-products each party may operate alone; purpose-built or fused data needs the others' consent); and parallel holders share a *joint defensive* interest against third parties. But the substance, he shows, falls back on derivative data — and here Xiong, Xu Ke (许可), and Shen Weixing (申卫星), despite different scenarios and tests, all tilt the derivative-data right to the *processor*, leaving the data contributor with contract/compensation/tort/PI remedies rather than ownership of the new product. DCC's read for overseas counsel: parallel property rights cut *attribution* uncertainty (who may use, operate, defend) but not *control* uncertainty (future use, detection, tracing, modelled value, third-party chains, ongoing compliance) — status, not control.
- § 02 · DATA-PROPERTY-RIGHTS
Why Upstream Won't Operate Its Data — Control Degradation, Derivative Data, and Irreducible Uncertainty
Part three of Hong Yanqing's (洪延青, 网安寻路人) study notes on China's 'separation of three rights' framework turns to the Right to Operate Data (数据经营权) — the right to provide data externally by transfer, licence, capital contribution, or pledge — and asks a question prior to 'what does operation transfer?': in real conditions, *will* an upstream party operate its data at all? His answer: yes, but narrowly. Control-dependent upstreams (platforms, holders of core user or irreplaceable industrial/training data) tend not to provide open, raw, autonomous access, and shift to controlled use or simply decline. The reason is structural. Once a downstream party is licensed to use data, the derivative data it produces is a *new object*: the upstream's *erga omnes* (对世) control over the raw data does not reach it, leaving the upstream — at most — a contractual claim against one counterparty. Hong then catalogues the uncertainties an upstream faces *ex ante*: some that attribution rules could touch but can't eliminate (qualification of the output, default ownership, good-faith of the processor, measurement of remedy), and some no rule can reach (combinatorial/unforeseeable value, undetectable misuse, the privity-and-insolvency chain, fusion and co-ownership, abstraction leakage into model parameters and learned skills, personal-information exposure, and counterparty hold-up). DCC's read for overseas counsel: this is the rigorous explanation of why Chinese data 'supply' is thin and why sandbox / privacy-computing structures dominate — defining a right does not supply the conditions to exercise it.
- § 03 · DATA-PROPERTY-RIGHTS
When the 'Right to Use Data' Goes External — Provision, Derivative Data, and the Erosion of Upstream Control
Part two of Hong Yanqing's (洪延青, 网安寻路人) study notes on China's 'separation of three rights' data-property framework turns to the Right to Use Data (数据使用权). The official definition (国家数据局, Common Data Terms Batch 2) makes the use right an *internal* power — 'I use my own data' to process, aggregate, analyse, and form derivative data — exercised on the premise of *not* providing data externally. So 'granting a use right to a downstream party' is not the use right travelling outward; it is the upstream party exercising its **operation right** to license, while the downstream party acquires a use right. That externalisation flips the downstream's legal position from PIPL **entrusted processor** (委托处理) to **provision** (提供) or **joint processing** — triggering notice and *separate consent* for personal information, and the Network Data Security Regulation's contracting duties. And because a strong use right lets the downstream form **derivative data** (衍生数据) — models, scores, indices, labels — value migrates downstream even though the raw data stays upstream. DCC's read for overseas counsel: in China data deals the use right is real but never self-bounding; whether a partner will grant an open, autonomous use right depends on its business model (control-dependent vs monetisation), and the default structure you should expect is *controlled use* (sandbox, privacy computing, federated modelling), not a clean copy.
- § 04 · DATA-PROPERTY-RIGHTS
Two Paths for the 'Right to Hold Data' — and Why the Narrow One May Add Little
Hong Yanqing (洪延青, 网安寻路人) works through the most unstable concept in China's 'separation of three rights' data-property framework — the Right to Hold Data (数据持有权). He pushes two readings to their logical ends. Path 1, the official 'complete separation' (三权完全切割): if the rights to hold, use, and operate data are truly independent, the holding right shrinks to a bare 'lawful-control state' whose only content is defensive — and that defense is already provided, against the world, by PIPL Article 10, DSL Article 32, the Network Data Security Regulation, and Article 13 of the Anti-Unfair Competition Law, so its incremental value as a standalone property right is thin. Path 2, the 'mother-right' reconstruction (持有权母权化): redefine 'holding' from factual control to a normative control that contains utilization potential, so the rights to use and operate are carved out from within it. DCC's read for overseas counsel: in Chinese data deals the tradeable substance sits in the rights to use and operate plus contract, registration, and compliance — not in 'who holds the data' — and China's data-property theory is still genuinely unsettled.
- § 05 · AI-GOVERNANCE
Prompt Stacks and Prompt Governance — Why System-Level Prompts Are Emerging as a Regulatory Lever (and Where They Fall Short)
A Chinese AI-law reading of Neumann, Sargeant and Singh's FAccT 2026 paper Prompt Governance? — and what it means for how China, the EU, and the US treat 'system prompts' as a regulatory object. Li Wenlong (科技利维坦) walks through the four-layer 'prompt stack' (system instructions → system guidelines → developer instructions → user prompts), five properties practitioners need to understand (layered, hidden, natural-language, malleable, loosely coupled to behaviour), and the comparative regulatory landscape: the EU GPAI Code of Practice requires signatories to disclose system prompts to regulators in model reports; the Trump EO 14319 / OMB M-26-04 stops at model / system / data cards and leaves system-prompt disclosure voluntary; the UK's AI Cybersecurity Code says effectively nothing. China's current GenAI safety regime (TC260-003 plus the GenAI Interim Measures) is output-evaluation-based — filing and pre-launch scoring, with no architectural hook into system prompts. Li predicts a Brussels Effect: system-prompt disclosure to regulators will become a global compliance baseline, analogous to the DPIA in data law. For overseas counsel: this is what is coming, what to start archiving now, and why 'what you write' in a system prompt is not 'what the model executes.'
- § 06 · AI-GOVERNANCE
Where China's Draft AI Anthropomorphic-Interaction Measures Need Work — A Scholar's Reform Map
Li Wenlong (科技利维坦) walks through the directions in which he would amend China's draft Interim Measures for the Administration of AI Anthropomorphic Interaction Services (人工智能拟人化互动服务管理办法) — the country's first dedicated rule on 'companion'-style AI. His critique is structural, not cosmetic: the core definition of '拟人化 (anthropomorphisation)' is too broad because it anchors on human-like expression rather than the real harm (relational dependency); the invented concept of '交互数据 (interaction data)' should be deleted and folded back into PIPL rather than blanket-prohibited; Chapter 2 mixes three incompatible duty types and should be split; the '1M registered / 100k MAU' security-assessment trigger is borrowed from other regimes and does not track real risk; and the training-data duties are horizontal obligations misplaced in a vertical rule. For overseas counsel building companion-AI or emotional-AI products for the China market: this is a map of where the draft is likely to move, and which duties fall on deployers versus base-model providers.