Why in News?
The Delhi High Court refused to grant interim relief to ANI in its copyright infringement case against OpenAI, holding that AI training on copyrighted content is covered by the fair dealing exception under the Copyright Act.
Key Highlights
- Justice Amit Bansal held that using ANI’s content to train Large Language Models (LLMs) falls under Section 52(1)(a) of the Copyright Act, 1957.
- The Court found that ANI failed to prove any loss to its news syndication business.
- ANI had earlier offered OpenAI a content licence for $7.5 million, indicating any potential damages are quantifiable.
- The Court rejected ANI’s request to delete training data, noting OpenAI’s legal obligations under US law.
- It observed that requiring licences from every copyright holder would make AI development economically unviable and highlighted AI’s public benefits in sectors such as education, healthcare, agriculture, and finance.
- The Federation of Indian Publishers, Digital News Publishers Association, and Indian Music Industry joined the suit.
Fair Dealing (Section 52, Copyright Act, 1957)
- A statutory exception permitting limited use of copyrighted works without the owner’s permission.
- Applies to purposes such as:
- Private or personal use (including research)
- Criticism or review
- Reporting current events
Copyright vs Patent
- Copyright: Protects original literary, artistic, musical, dramatic works, films, and software.
- Patent: Protects new inventions, granting exclusive rights generally for 20 years.
Copyright Act, 1957
- Governs protection of literary, artistic, musical, dramatic works, films, sound recordings, and software.
- Grants creators exclusive rights over reproduction, adaptation, distribution, and communication of their works.
- Section 52 provides exceptions under the doctrine of fair dealing.
PYQ (2014, GS3, 12.5 Marks) In a globalised world, intellectual property rights assume significance and are a source of litigation. Broadly distinguish between the terms copyrights, patents and trade secrets.
[2026] Which of the following statements with regard to Large Language Models (LLMs) used in machine learning is/are correct?
1. LLMs assign probabilities to the next possible words and then pick the one with the highest probability.
2. LLMs process data through mathematical optimization to minimise prediction errors.
3. LLMs produce unbiased outputs.
Select the answer using the code given below :
[A] 1 only
[B] 1 and 2 only
[C] 2 and 3 only
[D] 1, 2 and 3