The policy proposal, One Nation One License One Payment proceeds to a drastic re-tuning of an age of generative artificial intelligence of copyright governance. The heart of it is a doctrinal shift; one grounded in consent exclusivity is substituted with liability rule, supported by royalty, to have works that were copyrighted be used in AI training. Copyright owners have a right to receive compensation but cannot veto the use of training. It is therefore a proposal of transforming copyright into a right of exclusion to a right of compensation, thus, attempting to maintain economic value yet overcome failures in enforcement due to scale, opacity, and automation.
This intervention will be situated under two assumptions. First, AI training is not a legal nullity. Ingestion, storage and functional reuse of copyrighted works squarely brings under reproduction and adaptation rights of Section 14 of the Copyright Act, 1957, and cannot be casually caught up in fair dealing under Section 52. Second, conventional consent-based enforcement does not work structurally in the AI sense. Section 30 and 55 individual licensing is inappropriate to large-scale cross-border automated training pipelines, setting up a regulatory vacuum in which infringement is common but in effect impossible to challenge. The proposal reacts by transferring regulation, spanning injunctions to remunerations. This has two fault lines. One being royalty as regulatory point of reference and second being the displacement of the right not to give consent.
Royalty as the ‘Regulatory Anchor’
The main argument in proposal is that royalty (and not consent or injunction) should regulate AI training applications. The training of AI is regarded as a legally cognisable exercise of exclusive rights under Section 14 yet exclusionary remedies cannot be used. Royalty then serves the purpose of practical replacement of control, maintaining the economic core of copyright under which exclusion is a mere fiction. The Indian copyright doctrine offers spine to this move. Rulings have always been that reproduction is a use of action, but does not depend on dissemination. When AI training is defined as reproduction at scale, a remuneration right ensues as a matter of principle even in the absence of satisfying the traditional substantial similarity test in Section 51 of downstream outputs.
In the case Eastern Book Company v D.B. Modak, Supreme Court highlighted copyright is a statutory bundle of exclusive economic rights which are based upon skill and labour of author. Infringement of the normative basis of copyright through commercial appropriation of that labour without compensation is offensive. Similarly, India TV v Yashraj Films and RG Anand v Delux Films makes it clear that infringement analysis does not only focus on literal copying, non-literal and fragmented extraction would be enough in circumstances where value is being extracted out of the expressive institution. These principles compromise arguments that AI training, since it is non-expressive at the output layer, is royalty-free.
What is more important is the fact that Indian law does not lack the familiarity with liability-rule regimes. Already statutory and compulsory licensing regimes allow use without prior consent, in circumstances where transaction costs, diffuse ownership or refusal to license would inhibit access or innovation, as long as compensation is secured. AI training structure is structurally similar to a market failure with prohibitive transaction costs, asymmetrical bargaining power, and endogenous under-enforcement. It is against this context that, royalty based regime is not an aberrant but a doctrinal extension. Enforcement asymmetry is also taken care of by a centralised collective mechanism.
Individual creators can have no legal means of identifying or suing against unauthorised training, but a royalty system under statute will turn the pervasive under-compensation to the predictable reparation. However, there is a cost to raising royalty to be the main regulator reaction. Traditional Indian jurisprudence has always viewed copyright as a negative right, one of the power not to be used, rather than a claim to payment. The framework may undermine the normative foundation of the copyright by ensuring the right to access under the condition of payment. Royalty also presupposes that the right to a copyright is used up by economic compensation. This ignores non-price objections: reputation loss, creative out-innovation, being driven out of the market by systems that are trained on the works of an individual author. An equal, centrally fixed, and uniform royalty also stands the danger of under-rewarding high-value or labour-intensive works, reducing heterogeneity to administrative convenience. Tips Industries v Wynk Music warning about use of payment as a universal solvent. The Bombay High Court clarified that there cannot be an intent of monetisation as waiver of control, exclusivity and remuneration are not default substitutes.
An acceptable royalty regime should, then, be distinguished and managed. Flattening would be reduced through tiered rates which are regulated by the nature of work, magnitude of work and the nature of deployment. Court or tribunal control of the rate-setting would put the structure in line with the contemporary statutory licence jurisprudence, which would make it proportional, not bureaucratically obsessed. The architecture would be rounded off by transparency-lite obligations, which are adequate to check use without necessarily disclosing the entire datasets.
Right to withhold consent
The aspect of the proposal which is most disputable concerns removal of the right of copyright owner not to give his consent to AI training. This is a change of a property-rule regime whereby control is achieved by consent and injunctions to a liability-rule regime where access is ensured and disputes limited to compensation. The empirical argument to this change is simple. The training of AI is technologically opaque, diffusive geographically and ex ante resistant to policing. A veto right which is unenforceable, is liable to become normatively empty. Through substituting imaginary right of refusal with a certain compensation, the framework aims at saving economic worth of copyright. Indian law does not consider consent sacrosanct. The refusal has already been overridden by statutory licences in cases where intervention to protect public interest or market failure is required. Applying this reason to training of AI would be a continuum and not a discontinuity, assuming that the loss of power of veto is complemented by safeguards.
The counter-argument is more basic. Consent is not procedural in nature only but constitutes an authorial independence and command over the situations of exploitation. With support from Indian courts, payment has been established to replace authorisation on several occasions. The act of reproduction without consent was intrinsically unlawful in case of EM Forster v AN Parasuram. India TV v Yashraj Films repeats that, in all other areas other than the Section 52, consent is the requirement of the threshold. Tips v Wynk categorically denies any form of implied coercion to license just because they are being paid. More importantly, AI training is an extractive use, which is upstream. Training extracts unlike broadcasting or access-to-works regimes focused on propagation generate expressive value to create systems that can potentially directly replace creative labour. To creators whose markets AI output has structurally replaced, lack of ability to reject training can be detrimental and might be incurable through compensation. It also has constitutional undertones. Copyright has been identified as a proprietary interest; deprivation of control under Article 300A must meet the test of legality and proportionality. The broad withdrawal of permission is dangerous as it would transform the copyright into a controlled right that does not have proper customization.
A conditional consent-displacement model provides a more sensible compromise. Instead of abolishing refusal, reform ought to still allow consent to high risk applications, like training model which seeks to replace creative labour, without allowing liability-rule access in less risky situations. Opt-outs by category of sensitive classes of works, increased moral rights protection, and time-limited sunset provisions would maintain a fundamental area of authorial control, and deal with enforcement realities.
Situating India Globally
The proposed royalty-based approach in India is relatively in the middle ground. The US still uses ex post fair-use adjudication, and the US Copyright Office 2025 Report denies that AI should have blanket exceptions but indicates an apprehension about harms to the market and loss of license. The text-and-data-mining exceptions under the European Union protect text-and-data-mining with formal consent by opt-out, but there is no effectual enforcement of it as soon as the data is ingested. Japan TDM regime which is innovation first is becoming more and more criticised with the marginalisation of creators. Aborted opt-out plans in the UK demonstrate political opposition to permission wholesale abolition. It is against this background that the debate about India portrays a larger international reckoning, determining whether AI training can remain economically neutral or whether copyright needs to be structurally readjusted to inevitable technological utilization.
Authors’ Dictum: ‘Copyright after Control’
The One Nation, One License, One Payment scheme is not only a scheme of licensing proposal, it is also the re-conceptualisation of the operating logic of copyright. Its bet is that in an age when the technologically eroded control can be replaced with economic participation. Such a bet is justifiable, only when royalty is measured, assent is not destroyed, and interest of the authors more than price is institutionally acknowledged. When adopted as crudely and simply as a flat rate, mandatory access system, the plan threatens to drain the normative heart out of copyright. When adopted as a conditional, differentiated, and supervised system of liability-rule, it can be an effective re-engineering of copyright to the AI training economy, in which value can be preserved where control has become structurally impossible. It is not a matter of either innovation or ownership. It lies between considered refocusing and dogmatic dissolution.
(This post has been authored by Tanya Verma and Raman Singh Chauhan, 5th-year B.A., LL.B. student at Dr. Ram Manohar Lohiya National Law University)
CITE AS: Tanya Verma and Raman Singh Chauhan, ‘One Nation, One Licence, One Payment: Re-Engineering Copyright For AI Training Economy’ (The Contemporary Law Forum, 8 February 2026) <https://tclf.in/2026/02/08/one-nation-one-licence-one-payment-re-engineering-copyright-for-ai-training-economy/> date of access.