Why in the News?
Recently, the government released a working paper on Artificial Intelligence (AI) and Copyright Issues.
More on the News
- The paper has been prepared by a Committee constituted by the Department for Promotion of Industry and Internal Trade (works under the Ministry of Commerce and Industry).
- The objective of the Committee was to propose a framework that safeguards the rights of content creators while enabling responsible Generative AI innovation and equitable access to technology.
- There are 2 primary legal concerns concerning AI and Copyright:
- First, the use of copyrighted material as input for AI Model training;
- And second, the copyrightability of works generated by AI models.
- The working Paper focuses on aspect of Use of copyrighted material as input for AI training.
- The issues concerning the copyright status of AI-generated outputs will be addressed in future Working Paper.
Current Legal Framework in India related to Copyright and related issues
|
Key Issues & Arguments with Use of copyrighted materials as Input for AI Training
- Infringement of copyright Licences: Copyrighted content is often used in AI training without the necessary licenses from the rights holders.
- E.g. Ongoing litigation in ANI Media v OpenAI case revolves around whether OpenAI's use of news articles, literary works, and other content for training its LLMs constitutes copyright infringement, or whether such use can be categorised as 'fair dealing' under section 52 of the Indian Copyright Act 1957.
- Argument of Right Holders:
- Unregulated use of copyright-protected content may devalue human creativity and lead to underproduction of human-created content.
- Content industries are demanding that use of content for training of AI Systems should be subject to consent and compensation.
- Studies show AI models can memorise and reproduce copyrighted content, blurring the line between training and infringement.
- Copyright claims over AI-generated output:
- Whether AI-generated outputs are eligible for copyright protection and who should be recognised as the lawful author of such content.
- Attribution of liability in the event of Copyright infringement.
- Argument of AI System Developers:
- Access to extensive and diverse materials is necessary to develop effective AI Systems, reduce hallucinations, and mitigate bias.
- Excessive restrictions can hamper AI innovation.
- Training AI and creating datasets do not infringe on copyright because they focus on non-expressive elements of creative works.
Proposed Policy Framework by Report: The Hybrid Model (One Nation One License One Payment)
- Universal Blanket License: All lawfully accessed copyrighted works can be used for AI training under a single licence.
- Centralized Collecting Entity (CRCAT): A single, non-profit umbrella organization (Copyright Royalties Collective for AI Training) will collect royalties from AI developers.
- Flat Royalty Rates: Rates will be set by a government-appointed committee as a percentage of the AI system's gross revenue. This eliminates upfront costs for startups.
- Statutory remuneration rights: For the creators and copyright holders.
- The rights holders will not have the option to withhold their works for use in the training of AI Systems.
- Distribution Mechanism: Royalties are distributed to Registrants (creators who list their works in a "Works Database") via Copyright Societies or CMOs.
Other Existing AI Regulatory Models
|
Conclusion
The working paper seeks to strike a balance between promoting AI innovation and protecting creators' rights in the absence of clear copyright rules for AI. By proposing the "One Nation, One License, One Payment" hybrid model, it aims to ensure easy data access for AI developers while guaranteeing fair compensation and legal certainty for copyright holders.
