Whalesbook
India Proposes World's Most Ambitious AI Copyright Licensing Framework
India is poised to introduce what could be the most comprehensive regulatory move globally concerning the use of copyrighted material by Artificial Intelligence (AI) companies. A working paper from the Department for Promotion of Industry and Internal Trade (DPIIT) has put forth a novel licensing regime, aiming to address the contentious issue of AI training data.
The proposal suggests a significant overhaul of India's copyright laws, creating a framework where AI developers can access vast amounts of copyrighted content for training their large language models (LLMs) and frontier models. This initiative seeks to strike a balance between fostering technological innovation and protecting the rights of content creators.
The Core Issue: A New Licensing Regime
The DPIIT's recommendations center on a unique hybrid model. It proposes a single, 'blanket licence' that would grant AI developers universal access to virtually any copyrighted material for training purposes. This access, however, comes with a commitment to pay royalties to copyright holders. The royalty payments are structured to occur post-revenue generation, meaning developers would pay based on pre-determined rates once their AI models begin to earn income.
To manage this complex system, the proposal includes the establishment of a centralized body. This entity would be responsible for the intricate task of collecting and distributing royalty payments to the rightful copyright holders, ensuring a streamlined process for both AI firms and creators.
Concerns Over Creator Rights and Control
Despite the government's intent to mediate disputes between media houses, studios, and AI giants, the proposal has ignited significant debate among legal experts and creators. A primary concern is the lack of control afforded to copyright holders. Unlike regulations in regions such as the United Kingdom and the European Union, where creators can opt out of having their work used, India's proposed blanket licence offers no such mechanism.
Hemant Krishna, a partner at Shardul Amarchand Mangaldas & Co, highlighted the absence of gate access, stating that any AI developer willing to pay the applicable tariff could use a creator's work. Abhinav Shrivastava, a partner at LawNK Partners, echoed this sentiment, arguing that the current proposal could strip creators of their fundamental moral right to protect the integrity and integrity of their intellectual property. This lack of control extends to modification or adaptation, which is common in other licensing agreements.
The Compensation Puzzle
Beyond control, the proposed compensation mechanism faces substantial criticism. The reliance on post-revenue royalties, while seemingly equitable, presents significant practical challenges. Nikhil Narendran, a partner at Trilegal, pointed out the difficulty in tracing the specific contribution of a copyrighted piece of data to the final output of a complex AI model. This contrasts sharply with simpler models like music streaming, where revenue can be tracked per play.
Experts like Hemant Krishna suggest that this lack of clear attribution could lead to inequitable compensation. For instance, publishers with high-quality, valuable content might not receive proportionate royalty payouts compared to those publishing amateur or lower-value work, simply because the blanket system does not differentiate based on the value provided to the AI model. This could inadvertently incentivize the creation and submission of low-quality content ('slop') solely to maximize royalty collection, irrespective of its actual utility for AI development.
Seeking a Balanced Approach
While acknowledging the proposal's flaws, legal experts credit the DPIIT for recognizing the strategic importance of AI regulation and for attempting to create a 'sandbox' for AI development. The absence of a legal framework could indeed pose risks to intellectual property and investments in the AI sector.
However, many believe a simpler solution might be more effective. Nikhil Narendran advocates for monitoring AI-generated output for copyright infringement directly, bypassing the need for a complex royalty collection and distribution apparatus. Abhinav Shrivastava emphasized the necessity for extensive consultation with both the industry and the creator economy to refine the proposals and achieve a workable balance.
Ultimately, the debate underscores the challenge of applying existing IP frameworks, designed for static works, to continuously learning AI systems. India's proposed blanket license attempts a middle ground, but its success hinges on developing clear mechanisms for tracking usage, attributing value, and preventing the exploitation of low-quality content.
Impact
This proposal could significantly shape the future of AI development and the creator economy in India. It may lead to increased costs for AI companies, potentially impacting the speed of innovation, while also providing a much-needed revenue stream for content creators. The success or failure of this model could influence global AI regulation debates. Impact Rating: 9/10
Difficult Terms Explained
- Artificial Intelligence (AI): Technology that enables computer systems to perform tasks typically requiring human intelligence, such as learning, problem-solving, and decision-making.
- Large Language Models (LLMs): A type of AI specifically trained on vast amounts of text data to understand, generate, and process human language.
- Frontier Models: Advanced AI models representing the cutting edge of AI capabilities, often capable of performing a wide range of tasks with high sophistication.
- Copyrighted Content: Original works of authorship (like text, images, music) that are legally protected, granting exclusive rights to the creator.
- Licensing Regime: A system of rules and permissions that governs the use of intellectual property, often involving fees or royalties.
- Blanket Licence: A single license that grants permission to use a collection of copyrighted works without needing individual permissions for each.
- Royalty: A payment made to the owner of a copyright or patent for the use of their protected work.
- Department for Promotion of Industry and Internal Trade (DPIIT): A government department in India responsible for promoting industrial development and trade policies.