AI Large Language Models and Copyright Framework
The government working paper released on December 8, 2025, explores the relationship between AI large language models (LLMs) like ChatGPT and access to online content. The paper suggests that:
- LLMs should have access to freely available online content by default.
- There should not be an opt-out mechanism for publishers regarding content access.
- A non-profit, similar to a copyright society, is proposed to collect royalties for content use.
Key Stakeholder Opinions
- Nasscom's Dissent:
- Argues forced royalties equate to a "tax on innovation."
- Supports mining freely available content without paywalls.
- Advocates for content providers to reserve content from being mined.
- Committee's Response:
- Rejected Nasscom's argument.
- Highlighted small content creators' inability to enforce opt-outs.
Legal and Industry Implications
The Digital News Publishers Association has legally challenged ChatGPT creator OpenAI over copyright issues, which OpenAI denies. The working paper suggests not delaying decisions for litigation outcomes.
Proposed Royalty Framework
- Similar to India's "compulsory licensing" for radio stations.
- Access allowed with a statutory fee to rightsholders.
- Distribution of royalties would consider factors like web traffic and publisher reputation.
- Any decision can be appealed to the judiciary.
This framework aims to balance the interests of AI developers and content creators, despite potential resistance due to increased costs and valuation issues of content contributions.