DH ILLUSTRATION
The working paper published by the Department for Promotion of Industry and Internal Trade (DPIIT) committee (which appears to lack representation from content creators that hold copyrights, as well as AI developers) aims to tackle copyright concerns arising from generative AI, specifically focusing on compensating creators.
While the goal is noble, the heavy-handed approach threatens to stall innovation and ultimately fails the very creators it seeks to protect. India should adopt a market-driven approach that incorporates a text and data mining (TDM) opt-out mechanism for publicly accessible data and support commercial licensing agreements for proprietary datasets.
The DPIIT report is ambitiously titled ‘One Nation One Licence One Payment’. It proposes a mandatory blanket licence that ensures that no Indian copyright holder can withhold their data from being used to train AI. Under this framework, a statutory remuneration for creators is proposed to be established. In practice, this involves setting up a Copyright Royalties Collective that would collect payments from AI companies and distribute them to creators. This centralisation is fraught with challenges.
The core issue lies with how AI models (including large language models) work. They are probabilistic engines, not databases. They work by predicting tokens based on billions of parameters tuned during training. It is impossible to say with certainty if, and by how much, a specific work contributed to a specific output. The task is comparable to determining which grain of sugar made a specific bite of Mysore Pak sweet.
By mandating licence requirements, the policy misses the main point of copyright. The goal of copyright law is to promote human creativity by offering protection, so that “copycats” cannot steal with impunity. However, AI models are not copycats – while data is used for training, AI models seldom regurgitate the training data and therefore, are not direct competitors and do not necessarily reduce the market for specific pieces of copyrighted content. While AI-generated content competes as a whole with human content, it is non-specific to a particular creator, unless there is style mimicking.
There is also a problem of attributability, when similar content from multiple sources is part of the training data. Larger aggregators will eventually dominate small, high-quality creators due to the volume of their content. The report also admits this difficulty in attributability and proposes a flat rate as a proportion of global revenue. However, how can a fair rate be arrived at that represents the value of Indian datasets in the overall training data, and who gets what share of the revenue? These are very difficult issues to address.
The idea that a single government-mandated body can determine an appropriate royalty rate (for AI model developers) and a revenue split (for content creators) goes against the core economic tenet of price discovery through the market mechanism that can produce high value for all stakeholders. Policies must be judged on outcomes, not intentions. On that front, this policy fares rather poorly.
The report aims to make it easier for startups and small companies to build AI models by mandating blanket licences and charging royalties based on a fraction of global revenue. This is bound to fail for many reasons.
First, some AI model developers might choose to exit the Indian market, leading to less competition in the market. Many of them are burning cash without a viable path to profitability. The proposed framework also requires developers to disclose training datasets in detail. Barring a few open AI models, companies rarely disclose this data, as it is an ingredient of their secret recipe. Forcing these measures would slow the pace of innovation and adoption in India to a suboptimal state for creators, developers, and users alike.
Second, global AI model developers may choose not to train on Indian datasets with such a mandatory licensing arrangement. This creates a scenario where AI systems are powerful generally, but perform poorly when addressing India-specific issues or understanding Indian cultural or social nuances. We risk importing models that don’t understand us.
Third, smaller, high-quality content creators still receive the short end of the stick. Large aggregators will fill up the licence pool, ensuring they capture the bulk of the royalties. We see an example in the music streaming industry, where royalties from streaming providers fail to add up to a living wage for most artists.
Lastly, India stands to make significant productivity gains through the diffusion of AI. These benefits come from building applications that solve for India, even if the underlying models are not Indian. This is a gradual process where AI applications are adopted across sectors of the economy. In a bid to address compensation for creators, competition across the ecosystem and the pace of adoption of downstream applications will suffer.
The alternative
Technology platforms have disrupted conventional distribution and monetisation channels for content; generative AI further complicates it. The creator economy is facing a crisis globally and will need to rethink business models. The solution for creators might lie in increasing meaningful direct engagement with their consumers rather than obscure government-mediated royalties. Kevin Kelly’s concept of “1,000 True Fans”, each of whom is willing to pay a subscription fee, is one such pathway. Discoverability on AI platforms will be key to this, and creators will want AI to find them.
Nasscom’s dissenting comments in the report provide a reasonable alternate pathway. “For content that is publicly accessible online, rights-holders should be able to reserve their works from TDM through a machine-readable opt-out, at the point of availability. For content that is not publicly accessible, rights-holders should be able to reserve their works from TDM through contract or licence terms”.
Regulation is still evolving globally on this issue. It would be prudent not to shackle a nascent industry with heavy regulation.
(The writers are researchers with the Takshashila Institution)
Disclaimer: The views expressed above are the author's own. They do not necessarily reflect the views of DH.