Why in News?

• With the rapid rise of Artificial Intelligence (AI) and Large Language Models (LLMs) like ChatGPT, concerns have intensified over the use of copyrighted content for AI training without consent or remuneration.
• Globally, this has triggered litigation, policy debates, and regulatory uncertainty due to intersection of technology, IPR, innovation, and regulation; the role of the State in rate regulation and compulsory licensing.
• In this backdrop, a Department for Promotion of Industry and Internal Trade (DPIIT)-led committee has released a working paper proposing a statutory licensing framework to balance AI innovation with copyright protection in India.

What’s in Today’s Article?

• Key Proposal - ‘One Nation, One Licence, One Payment’
• Institutional Mechanism - CRCAT
• Royalty Determination Framework
• Retroactive Application of Royalties
• Transparency and Burden of Proof
• Stakeholder Responses
• Challenges in the Proposed Framework and Way Forward
• Conclusion

Key Proposal - ‘One Nation, One Licence, One Payment’:

• Mandatory blanket licence for AI training:
  • All AI developers must pay royalties for using copyrighted works in AI training.
  • No opt-out mechanism for freely available online content.
  • Model inspired by compulsory licensing in radio broadcasting under Indian copyright law.
• Rejection of voluntary licensing:
  • The committee rejects bilateral licensing deals (e.g., OpenAI–Associated Press).
  • Reasons are high transaction costs, unequal bargaining power, and marginalisation of small creators and startups.
  • Voluntary licensing is seen as favouring big tech and big publishers only.

Institutional Mechanism - CRCAT:

• A new umbrella non-profit body (Copyright Royalties Collective for AI Training [CRCAT]) to be established under the Copyright Act, 1957.
• Functions of the body include collecting royalties from AI companies, distributing proceeds among copyright holders, etc.
• Membership: Only organisations (not individuals), one member per class of work.
• Coverage can expand gradually to unorganised sectors.

Royalty Determination Framework:

• Government-appointed rate-setting committee:
  • Composition: Senior government officers, legal experts, economic and financial experts, AI and emerging technology experts, AI developers’ and CRCAT representatives.
  • Powers: Fix fair, transparent, predictable rates; review rates every three years; decisions subject to judicial review.
• Likely pricing model:
  • Flat rate preferred initially.
  • Royalty as a percentage of gross global revenue earned from commercialised AI systems (excluding taxes).

Retroactive Application of Royalties:

• Royalties to apply retrospectively: AI developers already using copyrighted works and earning revenue must pay past dues.
• Justification:
  • Ensures fairness and accountability.
  • Not punitive, but corrective to restore balance in the creative ecosystem.

Transparency and Burden of Proof:

• Data disclosure by AI developers:
  • Mandatory submission of a ‘Sufficiently Detailed Summary’ of datasets used.
  • Includes:
    • Type of data (text, image, music, audiovisual)
    • Source (social media, publications, libraries, public datasets, proprietary data)
    • Nature of data usage
• Distribution of royalties: CRCAT to distribute funds proportionally based on extent of usage, heavily used categories (news, music, audiovisual) receive higher shares.
• Legal presumption: In litigation, content owners claim it is presumed valid. Burden shifts to AI developers to disprove misuse or non-payment.

Stakeholder Responses:

• Supporters (Committee view):
  • Ensures non-discriminatory access to training data
  • Prevents concentration of royalties among a few big players
  • Creates a predictable legal environment for AI development
• Opponents:
  • NASSCOM:
    • Calls forced royalties a “tax on innovation”
    • Supports opt-out mechanisms for content creators
  • Creative industry concerns:
    • Government-fixed rates are globally unprecedented
    • Fear undervaluation of premium content

Challenges in the Proposed Framework and Way Forward:

• Risk of over-regulation stifling AI innovation: Ensure robust stakeholder consultation.
• Administrative complexity in royalty distribution: Fine-tune royalty rates to avoid discouraging AI startups.
• Resistance: From both AI firms (cost burden) and content creators (flat-rate concerns). Strong judicial oversight to prevent arbitrariness.
• India becoming a global outlier in AI copyright regulation: Harmonisation with global AI governance norms.

Conclusion:

• India’s proposed mandatory blanket licensing regime for AI training represents a bold and interventionist approach to reconciling innovation with copyright protection.
• By institutionalising royalty payments through a statutory mechanism, the Centre aims to ensure equitable compensation for creators while maintaining open access to training data for AI developers.
• The success of this model will ultimately depend on rate rationality, transparency, and adaptive governance, making it a critical test case for AI regulation in the Global South.