AI : Copyright: Legal Issue

Why in News? Whether the intellectual material produced by various generative AI models infringes copyright laws has been a controversial question posed around the globe.

Relevance : UPSC Pre &  Mains

Prelims :  Copyright related terms

Mains  : GS3

AI and Copyright: Legal Position – A Concise Summary

Do AI models violate copyright law?

• AI models are trained on vast datasets, including both copyrighted and public domain content. Courts have raised concerns about whether this training and the resulting outputs infringe reproduction rights or qualify as fair use (U.S.) or fall under text and data mining exceptions (EU, UK). The legality remains unsettled globally.

Can copyrighted content be used in training?
Recent U.S. rulings (2025)—Thomson Reuters vs. Ross Intelligence, Bartz vs. Anthropic, and Kadrey vs. Meta—suggest that training on legitimately acquired copyrighted data may qualify as fair use, provided:

• It’s transformative, not merely substitutive;
• It does not harm the original’s market.

However, training on pirated content or without regard for potential market impact remains legally risky.

What about AI outputs and ownership?

• There is no global clarity on who owns AI-generated content or whether it qualifies for copyright. Traditional IP laws, designed for human authors, do not directly recognise AI as a creator.

Legal concerns with databases:

• Use of proprietary databases for training raises issues under IP, contract law, and privacy regulations. While some countries offer fair use or data mining exceptions, there’s no harmonised international framework, leading to legal ambiguity.

 Indian context:

• Under the Copyright Act, 1957, creators have exclusive rights. Fair dealing (Section 52) may allow limited use, but AI-specific guidelines are lacking. India believes its existing IP framework suffices, though a clear stance on AI-generated content ownership is awaited. The ongoing ANI vs. OpenAI case may shape future interpretations.

Conclusion:
Courts are cautiously allowing transformative AI training under fair use, especially in the U.S., but pirated data, market harm, and lack of creator compensation remain legal and ethical concerns. India needs clearer AI-IPR norms aligned with emerging global standards.