Copyright Act 1957

The Indian government is undertaking a major reform of the Copyright Act, 1957 to address the challenges posed by Artificial Intelligence (AI) in the creation and use of copyrighted works. This reform is based on a hybrid model, balancing the needs of AI innovation with the rights of creators.

Hybrid Model Overview:

• Introduces a mandatory blanket license allowing AI developers to use copyrighted works for training AI systems.

• Grants creators a statutory right to remuneration for the use of their works.

• Aims to balance innovation and creator rights, positioning India differently from the European Union, which emphasizes transparency, and the United Kingdom, which provides limited recognition to computer-generated works.

Proposed Changes to the Copyright Act, 1957

1. Blanket License for AI Developers

• AI developers will receive a non-negotiable, universal license to use all lawfully accessed copyrighted works for AI training.

• Rights holders cannot withhold their works from being used in AI training.

• Eliminates the need for individual permissions currently required under Section 14, which grants the owner exclusive rights to reproduce, adapt, or communicate their works.

2. Statutory Remuneration Rights

• Copyright holders will be entitled to royalties even without providing explicit consent.

• Royalties will be calculated as a fixed percentage of revenue generated from AI systems trained on copyrighted content.

• AI firms pay royalties only after commercialization, not during data collection or training.

3. Centralized Collection and Distribution Mechanism

• The government will establish a centralized non-profit entity to collect royalty payments from AI developers.

• This entity will include Copyright Societies and Collective Management Organizations (CMOs) as members.

4. Protection for Startups and MSMEs

• The proposed reforms may include concessional provisions or differential royalty structures to support startups and MSMEs, ensuring a level playing field.

Need for Amendments

1. Absence of AI-Specific Definitions

• The current Copyright Act does not define terms such as Artificial Intelligence, Generative AI, machine learning, or training data.

• This creates legal uncertainty about how AI systems and text-and-data mining (TDM) activities fit under the existing law.

2. Ambiguity on AI Training as Copyright Use

• It is unclear whether AI training constitutes reproduction, adaptation, or infringement under Section 14.

• There is uncertainty about whether AI training qualifies as fair dealing under Section 52.

3. Unauthorized Use of Copyrighted Materials

• AI systems currently rely on mass-scale scraping of copyrighted content without consent, bypassing the permission-based framework of the Act.

4. Copyrightability of AI-Generated Works

• Section 2(d)(vi) defines an author as a human, leaving AI-generated content without clear ownership or copyright protection.

• Section 2(ffc) treats computers as tools, adding further ambiguity over AI-created works.

5. Cross-Border and Jurisdictional Issues

• Section 40 protects foreign works territorially but does not address AI systems trained abroad on Indian works or cloud-based AI models, creating enforcement challenges.

Key Sections of the Copyright Act, 1957

• Section 2: Defines types of works covered, e.g., literary works (2(o)) and dramatic works (2(h)).

• Section 13: Extends copyright protection to literary, musical, dramatic works, films, and sound recordings.

• Section 14: Grants exclusive rights to reproduce, adapt, publish, translate, or communicate works, which cannot be exercised without the owner’s permission.

Judicial Interpretations

• Mr. Dattatray Bapu Dighe v. State of Maharashtra (2024): Copyright exists automatically upon creation; registration is not mandatory.

• Star India Pvt. Ltd v. Magicwin.Games (2024): Digital piracy websites hosting copyrighted content were permanently blocked.

• India TV v. Yashraj Films (2012): Small usage of songs does not constitute copyright infringement.

• Oxford v. Rameshwari Photocopy Services (2016): Photocopying book excerpts for educational purposes is fair dealing, promoting public access to knowledge.

Global Approaches to AI-Generated Content

Conclusion

The proposed amendments to the Copyright Act, 1957 aim to:

• Introduce a mandatory blanket license for AI training.

• Ensure statutory royalties for creators.

• Address gaps in AI-related definitions, training, and authorship.

• Establish a centralized royalty collection mechanism, ensuring fair compensation while promoting AI innovation in India.