Copy article

India Considers AI Royalties: OpenAI and Google Should Pay for Local Training Data

ended 11. December 2025

India's Department for Promotion of Industry and Internal Trade (DPIIT) has published a 79-page working paper proposing mandatory royalties for AI companies using Indian copyrighted content in model training. The framework, titled "One Nation One License One Payment," establishes a centralised collection system designed to guarantee creator compensation while ensuring AI developers access to training data.

Mandatory Blanket License System:

  • AI developers receive automatic rights to use all lawfully accessed Indian copyrighted content for training
  • Copyright holders cannot withhold their works from AI training use
  • Royalties paid as percentage of revenue generated from AI systems trained on copyrighted content
  • Central government-designated non-profit entity collects and distributes payments

Copyright Royalties Collective for AI Training (CRCAT):

  • Centralised collection body comprising Copyright Societies and Collective Management Organisations
  • One member organisation per content class (text, music, images, audiovisual)
  • Distributes royalties to both members and non-members who register works
  • Government-appointed committee sets royalty rates rather than marketplace negotiation

The UK launched a consultation in December 2024 proposing a text and data mining (TDM) exception with opt-out rights for creators.  It received 11,500 responses with intense creative industry opposition (the "MAKE IT FAIR" campaign). By July 2025, rather than implementing the proposal, the government established expert working groups to find solutions, with the formal response to the consultation still awaited. 

The Indian committee has rejected opt-out mechanisms, describing them in the working paper as leaving "small creators largely unprotected owing to lack of awareness, bargaining power, and mechanisms to verify compliance" and criticising them for shifting "the burden from content users to content creators."

Nasscom, representing the the 3,000 strong membership of Information Technology (IT) and Business Process Management (BPM) firms in India, has formally opposed the mandatory licensing framework, warning that compulsory payments could "constrain innovation and disproportionately burden young companies." The industry body argues instead for a text-and-data-mining exception with an opt-out mechanism.

We want your views:

  • Can government committees set fair royalty rates without market price signals?
  • Does mandatory participation protect small creators or remove their agency?
  • How does CRCAT verify which works were used when companies resist disclosure?
  • Will lawful access requirements work when training crosses global jurisdictions?
  • What does this mean for the UK and other nations, who are all wrestling with a range of potential compensation models and are suffering analysis paralysis because of the AI arms race.

 

4 responses from the Newspage community

Copy all

Star Quote
Copy

Colette MasonVerified
AI Ethics Consultant at Clever Clogs AI
Revenue-based statutory rates mean The New York Times archive gets paid the same percentage as scrappy blog posts. The rate reflects AI company revenue, not content quality. Small creators gain protection without needing lawyers, but major publishers and quality producers lose the ability to negotiate premium licensing deals for exclusive content.

AI companies must submit dataset summaries, but there's no independent audit mechanism to verify self-reported data against actual training inputs. India proposes collecting royalties on global revenue from models trained anywhere on Indian content, but enforcing payment from offshore entities is the hard part.

If India's mandatory payment system works, it becomes a global template forcing AI firms to pay for content everywhere, finally ending the "scrape it free" era.

The question then becomes is the money allocated to the sources that deserve it, rather than participated, irrespective of information quality.
Copy

Editor
Mitali DeypurkaysthaVerified
AI Governance Strategist & CMO at AISensa
If creators had been paid from day one, AI would look completely different today. But this is what happens when profit outruns principle. Tech companies scraped the internet for years, telling creators it was “good for visibility,” and now those same works are being fed into AI models without consent or compensation. India’s move forces Big AI to finally pay, but laws only exist when companies prove they can’t be trusted to do the right thing.

The danger is that royalties may end up being pennies, just like music streaming: technically “payment,” but barely survival. The big players will capture most of the value while small creators fight for crumbs. India shouldn’t have needed this law. Ethical collaboration should have come first. But when tech treats creativity as free fuel, governments are forced to step in and create blunt laws. AI doesn’t need endless data extraction to grow; it needs a fair system that honours the humans who built the internet in the first place.
Copy

Editor
Kate UnderwoodVerified
Founder & Chief People Strategist at Kate Underwood HR and Training
Calling it protection while small creators lose all choice is not a fair deal; it is a polite land grab. Most small creators do not have the time, lawyers or tech skills to police how their work is used in AI training. So yes, a simple system that makes big firms pay for using their content could finally turn unpaid scraping into real income. That is the good bit. But if you force everyone in, with no real right to say no, you are treating their life’s work like a utility bill. Real protection means three things: clear info on who has used your work, easy ways to sign up and get paid, and a genuine “hands off this piece” option for sensitive or high-value stuff. If creators cannot ever opt out, it is not protection; it is compulsory discounting.
Copy

Rohit Parmar-MistryVerified
Founder at Pattrn Data
It is refreshing to see a government finally reject the ‘opt-out’ trap. The UK is stuck in analysis paralysis while India correctly identifies that opt-outs are useless for small creators who lack the legal budget to police OpenAI. Shifting the burden onto the trillion-dollar companies is the only practical path.

However, the execution is a minefield. A government committee setting royalty rates without market signals is a recipe for disaster. We are asking companies that call their models ‘black boxes’ to suddenly itemise exactly whose data generated profit.

The real loophole? Revenue-based royalties. Big Tech is world-class at creative accounting. If royalties are tied to direct revenue, expect AI models to be classified as ‘loss leaders’ or bundled into broader subscriptions so the specific AI revenue is invisible. Unless the fee is based on usage or compute rather than nebulous ‘revenue’, corporate lawyers will ensure the royalty pot remains empty.