Intellectual Property Rights in the Digital Era: Looking at Fair Dealing vs. Market Copying

Introduction and the Problem

Artificial Intelligence tools like ChatGPT and Midjourney have Completely transformed how society perceives modern technology. These tools cannot learn independently. They must read, copy and analyze millions of books, articles and pictures from the internet. Most of this data belongs to creators who never gave their permission. Under Section 14 of the Indian Copyright Act, 1957 only the original creator has the right to copy their work.¹ When an AI company downloads thousands of articles onto its servers without a license, it may constitute infringement under Section 51, unless a statutory exception under the Act applies.

To defend themselves AI companies use an excuse called "Fair Dealing" found in Section 52 of the Indian Copyright Act 1957.² The case from the Delhi High Court called ANI Media Pvt. Ltd. V. OpenAI Inc. is one of the well-known case.³ The media company claims OpenAI stole its news articles while the tech company argues it is just learning. The main question is: Is scraping data to train an AI a type of research or is it just a cheap way to replace human creators? This article examines whether AI training is fair under the Copyright Act 1957. AI companies are using copyrighted data to train their model without permission. It involves web crawling and web scraping.

Why Indian Law is Different and Strict

Many people confuse foreign law with Indian law. In the US judges use a rule called Fair Use to decide if copying is fine. In India we use a much stricter system called Fair Dealing. If your action does not fit the written statute, it is violation of the right. AI companies usually point to Section 52(1)(a)(i) of the Indian Copyright Act. This section says it is okay to use copyrighted work without permission, if it's for private or personal use, including research. The tech companies argue that an AI is simply doing computer research and look the pattern when it reads data. However, the final product is sold to the public for a profit. In the case Oxford University Press v. Rameshwari Photocopy Services (2016), the Delhi High Court said that the purpose of copying must perfectly match what the law allows.⁴ If these AI tools meant to make money, they do not qualify as private or personal research. AI companies are making money from AI Training by indirectly copying the artist's work.

The Argument Over Intermediate Copying

AI developers have another defense. They cite an Indian Supreme Court case called R.G. Anand v. Delux Films (1978).⁵ This case established that for something to be a copyright violation the final product must look like a copy of the original. Tech companies argue that an AI does not creates same like the original. Instead, it breaks the words down into numbers and probabilities to create a completely new sentence. However, this argument ignores the step of making an AI. Even if the output looks new, the input stage requires making a copy of the humans work onto the AI's computers. Indian courts do not overlook this step. In law, downloading a file or scraping a database without permission is infringement. The fact that the AI deletes the file later or changes it into new work does not wipe away the fact that they copied the original work.

Testing AI with the "Three Rules of Fairness"

To figure out if an action is truly fair, Indian judges often look at a landmark case from the Kerala High Court which is Civic Chandran v. Ammini Amma (1996).⁶ This case created three rules to test fairness:

1. How much of the work did you take?

2. Why did you take it?

3. Will your new product compete with the creator?

When we apply these rules to generative AI, the tech companies will lose on all three counts. First the AI takes 100% of the articles and books, not just a small quote. Second, the purpose is commercial, which means the tech companies want to make money. Third and most important, the AI directly competes with the creator. If a newspaper office can use an AI tool to write summaries in the style of a journalist, then they will stop hiring that journalist. It will be more cost effective to them. AI uses the creator's data to put that creator out of a job. This directly violates the fairness of the original creators.

Even though the law favors creators, India faces some difficulties. In Europe the law allows data mining. They let the creators opt-out using technical tags. In places like Japan, the government made a law that allows AI companies to scrape anything for free to boost their tech industry. If India makes its copyright laws too strict, it could hurt the technology sector. If they are constantly sued or forced to pay millions of rupees for licenses, Indian tech will fall behind America and Europe. This creates a conflict between protecting artists and supporting technology goals. We need to find a solution to this problem. We need to protect our artists and support our technology sector.

Another important issue is lack of transparency and consent. As there is no special technique to detect whether the creator's article or art is used, they didn't even know that their work has been used. So, AI companies should clearly record which work has been used to train their model and by determining that the government can decide which artists or creators should get the compensation for that. By using this way, it is not only supporting the innovation of the technology but also protect the rights of the creators.

Conclusion and a Smart Solution

In my view, under the existing principles of Indian law, scraping internet data for free commercial AI training is a violation of copyright. It may not fall within the statutory exceptions for fair dealing and may cause significant harm to the livelihoods of human creators. However, we cannot simply ban AI training if we want the advance technology. The best solution is to create a law called a "Statutory License." Under this system AI companies would have the right to scrape websites for training data without asking every single person for permission. In return, the AI companies would pay a fee into a government fund. This fund would then distribute the money back to writers, journalists and artists based on how much their work was used. This simple solution protects creativity while letting Indian technology grow fairly. AI Training can be fair and legal, in India.

References

Statute

1.The Copyright Act, 1957, No. 14 of 1957 (India).

Cases

2. ANI Media Pvt. Ltd. v. OpenAI Inc. & Anr., 2026 LiveLaw (Del) 687 (Delhi High Court 2026).

3. Civic Chandran v. C. Ammini Amma, 1996 (16) PTC 670 (Ker. H.C.).

4. R.G. Anand v. Delux Films, (1978) 4 SCC 118; AIR 1978 SC 1613.

5. The Chancellor, Masters & Scholars of the University of Oxford v. Rameshwari Photocopy Services, 2016 SCC OnLine Del 5451 (Delhi High Court 2016).

Secondary Source

6. Khurana & Khurana, "Web Scraping and AI Training Data: Copyright Challenges in the Age of Generative AI," Apr. 29.