Team Insights

Market Restructuring in the Era of Zero Copyright Protection for AI Creations

Rongchao IP Team · 2026-04-16

A Law-and-Economics Report from the Perspective of Law as a Public Service Product

I. Reconstructing the Theoretical System: Copyright Law as a Public Service Product in an Age of Near-Zero Marginal Cost

Before analysing how copyright markets would be reshaped if AI-generated content were denied copyright protection, it is necessary to return to the jurisprudential and economic nature of law itself. In traditional private-rights discourse, intellectual property is often misunderstood as a natural right akin to property in land or chattels. Modern jurisprudence and public economics point in a different direction. Law is a public service product supplied by the state, or by a transnational political entity. Its function is to stabilise expectations, reduce social transaction costs, correct market failure and maintain the public domain.

Copyright law is one branch of that public service. Its economic purpose is not to confer metaphysical ownership over expression, but to improve long-term social welfare by encouraging the production and consumption of creative, scientific and cultural works. Information goods have public-good features: they are difficult to exclude without legal compulsion, and one person’s consumption does not reduce another’s. This creates a familiar market failure. Authors incur high fixed costs in creating a work, while subsequent copiers face low fixed costs and marginal copying costs close to zero. Copyright supplies a time-limited statutory monopoly so that the author may price above marginal cost and recover fixed investment.

That monopoly is costly. It restricts public access to existing works, which are themselves inputs for future creation. Copyright therefore always rests on a fragile balance between private incentive and public knowledge diffusion.

Generative AI alters the premise of this model. It does not merely reduce the marginal cost of copying, as the internet did. It also reduces the marginal cost of creation itself. When a symphony, a commercial illustration or a detailed industry report can be generated within seconds at very low computing cost, the fixed-cost-recovery justification for copyright becomes less persuasive. If production cost is already low, copyright-like exclusivity may be suboptimal protection: it restricts access without being necessary to induce production.

Accordingly, if a legislature excludes purely AI-generated content from copyright protection, it is not confiscating a natural right. It is declining to provide a public service, namely exclusive proprietary protection, for outputs that no longer present the fixed-cost problem that copyright was designed to solve. The phenomenon may be described as res publicae ex machina: machine-generated public goods. In the short term, the release of large quantities of unprotected content may be a Pareto improvement. In the longer term, however, the absence of this public service will alter supply and demand, value-chain allocation, transaction-cost structures and the position of human creators.

II. Global Differences and Policy Convergence: A 2025-2026 Jurisdictional Overview

By 2026, major jurisdictions have begun to converge on a baseline rule: purely autonomous AI output is not protected by copyright. Jurisdictions differ on how much human intellectual input is sufficient, and on the evidential burden for enforcement, but the exclusion of purely machine-generated expression is becoming a cross-system consensus.

The United States, as both a major exporter of copyright works and a centre of AI development, has maintained the constitutional requirement of human authorship. The US Copyright Office’s 2025 report on copyrightability and AI restated that existing principles are flexible enough to address the technology, but only works in which human authors provide sufficient creative selection, arrangement, modification or dominant expressive contribution are protectable. A prompt, however detailed, does not by itself give the user sufficient control over the final expressive form. Federal decisions concerning Stephen Thaler’s AI-generated images have reinforced this position: extending protection to material whose core expressive elements are determined by a machine would not advance the constitutional purpose of copyright.

China has taken a more pragmatic and moderately tolerant approach. Courts have not absolutely excluded the possibility that AI-assisted images may attract copyright, but they have placed a heavy burden of proof on claimants. In the “cat crystal pendant” image case, the Beijing Internet Court required the claimant to explain the creative idea, produce the evolution of prompts, and preserve records of repeated selection, tuning and modification. A later simulation using the same software was not enough. Other courts, including the Zhangjiagang court in a furniture-design dispute, have rejected claims where human creative intervention was insufficiently evidenced. On platform liability, Chinese courts have also begun to explore filtering and blocking obligations for AI image-generation services, as reflected in disputes involving well-known character images.

The European Union’s position is more structural. EU copyright doctrine remains anchored in the author’s own intellectual creation, while the AI Act, transparency obligations and policy debates on generative AI point towards a division between protectable human expression and unprotected machine output. Japan’s approach is more flexible on training-data use, but copyright protection of outputs still depends on human creative contribution. Across these jurisdictions, the policy direction is clear: the public service of copyright protection is reserved for human creative agency, not for fully autonomous machine production.

III. Economic Effects in a No-Copyright Market: Supply-Demand Mismatch and Forced Reversal of the Value Chain

If AI-generated content is not protected, the first market effect is a dramatic increase in supply. Images, music, text, video concepts and advertising copy can be produced at scale. The abundance of content pushes the price of generic expression towards zero. The scarcity shifts away from the content unit itself and towards attention, distribution, verification, reputation and integrated service delivery.

The second effect is a reversal of value-chain bargaining power. In the traditional model, the owner of the copyright work controls licensing. In a no-copyright AI-output market, downstream users may freely copy and adapt the output unless another right, such as trade mark, passing off, unfair competition, privacy, contract or platform rule, intervenes. The legal rent that once attached to the work is displaced by business-model rents attached to access, certification, community, speed, data advantage and client-specific service.

The third effect is a mismatch between macro abundance and micro trust. The market may receive more content than it can verify. Buyers no longer ask only whether a file is available; they ask whether it is authentic, licensed, non-infringing in relation to inputs, commercially safe, reputationally suitable and traceable. A market saturated with unprotected AI output therefore creates demand for new legal and technical services: provenance verification, chain-of-title review, training-data warranties, indemnities, insurance and platform-level governance.

Human creators do not simply disappear in this model. Their economic position changes. They must compete less on the mere production of content and more on judgment, taste, authenticity, narrative authority, client relationships and legally reliable provenance. The premium moves from “I made a file” to “I can stand behind this work, its origin, its quality and its commercial use”.

IV. The Rise of Privately Customised Public Services: Interoperability and the Institutionalisation of the Authenticity Premium

Where the state withdraws copyright protection from purely AI-generated works, private actors respond by building substitute governance systems. These systems are not copyright, but they perform some of copyright’s coordination functions. They include platform certification, content credentials, provenance ledgers, model-use policies, contractual warranties, insurance products and authenticity labels.

Technical interoperability is central. Standards such as C2PA and content authenticity infrastructures allow a file to carry information about origin, editing history, software tools and publisher assertions. These mechanisms do not prove legal entitlement in the full sense, but they reduce information asymmetry and transaction costs. They create a market signal: verified human authorship, licensed training origin, or authorised AI assistance may command a premium over anonymous synthetic content.

This produces an authenticity premium. In fields such as journalism, legal analysis, education, cultural heritage, luxury goods, music and visual licensing, buyers may pay not merely for output but for trustworthy origin. Authenticity becomes a privately supplied public service. It is public in function because it stabilises the market and enables reliance; it is private in supply because platforms, agencies, collectives and certification bodies provide it through contracts and technology rather than through copyright registration alone.

V. Defensive Reorganisation of Traditional Copyright Industries: Platform Governance and the Authenticity Cartel

Traditional copyright industries will not passively accept the collapse of legal exclusivity for AI outputs. They are likely to reorganise defensively. Music platforms, image libraries and media distributors are already strengthening rules on AI labelling, impersonation, voice cloning, catalogue ingestion and unauthorised synthetic substitution.

The strategic objective is to protect scarcity. If AI output is abundant and unprotected, incumbent platforms can shift competition to authenticity and safety. Spotify’s AI-music policies and partnerships with labels point to a model in which AI tools are permitted, but only within an artist-first and rights-controlled environment. Getty Images and Shutterstock’s consolidation and their commercially safe AI offerings likewise suggest a move from licensing individual images to licensing trusted datasets, indemnified outputs and certified visual supply chains.

This may generate what can be called an authenticity cartel. The term does not necessarily mean an unlawful cartel in the antitrust sense. It describes a market structure in which a small number of platforms control the infrastructure for certifying what is “real”, “human”, “licensed” or “commercially safe”. The risk is that private authenticity governance may become exclusionary. Independent creators and small AI developers could be priced out of verification systems, while dominant platforms convert trust infrastructure into a barrier to entry.

VI. Alternative Monetisation: From Digital Rights Management to Reverse Paywalls

In a market where AI outputs are not protected by copyright, monetisation shifts from ownership of copies to control over access, verification and use context. Digital rights management, which historically tried to prevent copying, becomes less central. The more important question is whether an AI system, crawler, platform or enterprise user must pay to access high-quality human content, verified datasets or trusted archives.

This creates the logic of the reverse paywall. Instead of charging human readers to access content, publishers and creators charge AI systems, platforms and commercial users for lawful ingestion, training, retrieval or synthetic transformation. The right being monetised may be contractual rather than copyright-based. Access controls, database rights in some jurisdictions, anti-circumvention rules, trade secrets, unfair competition, API terms and technical measures become part of the monetisation package.

Blockchain and full-life-cycle recording tools may also be used to document creation, modification and licensing events. Their value is not that they magically create copyright in AI outputs, but that they produce evidence. In disputes over provenance, contractual scope, training-data compliance or misrepresentation, evidence may be the monetisable asset.

VII. Conclusion and Policy Implications: Rebuilding Public Services for the Age of Intelligent Machines

The exclusion of purely AI-generated content from copyright protection should not be viewed as a gap or failure in the law. It may be a deliberate recalibration of the public service supplied by copyright. Where the economic justification for exclusivity has weakened, the state may properly decline to grant monopoly protection.

That withdrawal does not mean legal disorder. It means that other institutions must absorb functions previously performed by copyright: provenance, authenticity, allocation of risk, platform governance, licensing of inputs, consumer protection and competition control. The future copyright market will therefore be less about the exclusive ownership of isolated outputs and more about the governance of production systems, verified supply chains and trustworthy human-machine collaboration.

For policymakers, the task is to avoid two errors. The first is over-protection: granting copyright to machine output without human creativity, thereby creating unnecessary monopolies over abundant content. The second is under-governance: leaving authenticity, provenance and platform power entirely to private ordering. A sound regime should preserve the public domain for machine-generated abundance while building enforceable rules for transparency, labelling, training-data compliance, personality rights, unfair competition and market access.

For creators and businesses, the practical message is equally clear. The legal value of content will increasingly depend on human contribution, evidence of process, contractual control of inputs, platform certification and commercial trust. The winners will not be those who merely generate more content, but those who can prove origin, manage risk and deliver reliable expressive products in a market flooded with synthetic abundance.