Mint Explainer: The OpenAI case and what’s at stake for AI and copyright in India
Summary
- A groundbreaking legal battle is unfolding in India with ANI filing a lawsuit against OpenAI, claiming copyright infringement over the use of its content to train AI models.
- This case raises critical questions about copyright and fair use, and could reshape the legal landscape for GenAI in India.
The first legal challenge to generative artificial intelligence models in India surfaced last week when news agency ANI filed a lawsuit against OpenAI in the Delhi High Court alleging copyright infringement. In this case, a first-of-its-kind in India, New Delhi-based ANI has accused the San Francisco-headquartered creator of ChatGPT of using ANI’s copyrighted content to train its language models. Mint reached out to experts to understand the stakes in this case and how it could determine India’s legal stance on GenAI.
ANI v OpenAI: What is the case?
On 19 November, ANI moved a copyright infringement lawsuit against OpenAI, accusing the company of using its copyrighted material to train its language models. The lawsuit, filed in the Delhi High Court, seeks damages of ₹2 crore and an injunction to prevent OpenAI from storing, publishing, reproducing, or using ANI’s copyrighted works.
ANI’s plea outlined three primary causes of action against OpenAI.
- It alleged that OpenAI used ANI’s publicly available copyrighted content to train its large language models (LLMs) without obtaining authorization.
- ANI cited instances where ChatGPT generated responses that were either verbatim or substantially similar to its copyrighted material.
- ANI raised concerns about “hallucinated" responses, where ChatGPT falsely attributed fabricated interviews or news stories to the agency. ANI argued that this misuse of its content not only infringed its copyright but also raised significant public interest concerns, particularly in the context of false political news that could have far-reaching consequences.
How has Open AI responded?
OpenAI has denied any wrongdoing, asserting that copyright laws protect the expression of ideas, not facts or ideas themselves. The company highlighted that ANI has not cited specific instances of its copyrighted material being reproduced by ChatGPT.
OpenAI informed the Delhi High Court that it had blocklisted ANI’s domain as of October 2024, ensuring that ANI’s material was no longer used for training its models.
The company also emphasized that its servers are located outside India, arguing that this makes the lawsuit jurisdictionally irrelevant. OpenAI also pointed out that none of the lawsuits it has faced globally, including in the US, Canada, and Germany, had resulted in injunctions or findings of copyright infringement.
A spokesperson for OpenAI, in a statement, said: “… we build our AI models using publicly available data, in a manner protected by fair use and related principles, and supported by long-standing and widely accepted legal precedents."
What has the court said so far?
A single bench of the Delhi High Court, led by Justice Amit Bansal, issued notices and summons to OpenAI but refused to grant an interim injunction in ANI’s favour.
The court acknowledged the case’s complexity as the first of its kind in India and underscored the need for detailed deliberations. The court has appointed an amicus curiae (an expert in the field to assist the court) and scheduled the next hearing for 28 January 2025.
What does this case mean for AI regulation in India?
Experts believe the outcome of this case could establish a legal framework for regulating AI content in India, particularly concerning copyright violations. If the court rules against OpenAI, it could set a precedent for stricter rules on using copyrighted content for training AI models.
“ANI’s lawsuit against OpenAI marks a pivotal moment for generative AI jurisprudence in India. The case underscores the growing tension between AI innovation and intellectual property protections," said Amit Panigrahi, partner at Luthra & Luthra Law Offices in India.
Advocate Ayush Jindal, founder of BeLegal, added that the case should establish guidelines for the ethical use of copyrighted content for training AI models. “It could also determine if Indian copyright law considers AI-generated outputs as derivative works," he said.
Also read | ANI vs OpenAI: Why ‘hallucinations’ are unlikely to go away soon
Even an interim judgment by the Delhi High Court will significantly influence AI deployment in India, added Ameet Datta, partner at law firm Saikrishna & Associates. “This case asks an Indian court to decide whether the unauthorized scraping and use of copyrighted content for AI training constitutes copyright infringement and whether AI outputs based on such data also infringe copyright."
Latha R. Nair, partner at K&S Partners, explained that Indian copyright law lacks exceptions for text and data mining, which is essential for machine learning, and this case could push for changes in the law.
What key AI and copyright-related issues can this case address?
As the first case of its kind in India, the ruling is expected to address several critical questions surrounding copyright and AI:
- Whether scraping and using copyrighted content without authorization for training AI models constitutes copyright infringement.
- Whether such use can be considered “fair dealing" under India’s Copyright Act, which allows exceptions for purposes like research, criticism, or reporting. This will test the boundaries of what is permissible within the existing copyright framework.
- Whether the concept of “transformative use", a principle recognized in US copyright law, can be applied in Indian jurisprudence. Transformative use typically allows for a defence when the original work is significantly altered or repurposed, raising important questions about its relevance to Indian fair use defences.
Swati Sharma, partner and head of intellectual property at Cyril Amarchand Mangaldas, explained: “The case will likely address whether downloading and copying copyrighted content for AI training constitutes infringement, whether AI models or their outputs can be considered derivative works, and if fair use can serve as a defence in such claims."
Ronil Goger, managing partner at Blaze Legal, added that legislation and court rulings must clarify the originality standards for AI-generated works, adding that the UK, the US, and Canada prioritise effort over novelty, acknowledging the collaborative role of human creativity in producing AI-driven content.
How can this case empower publishers?
For one, ANI’s case against OpenAI “is likely to open the floodgates for more publishers filing lawsuits for copyright violations", said Sharma of Cyril Amarchand Mangaldas. “Many may also consider intervening in the current proceedings."
Panigrahi of Luthra & Luthra added that the case outcome could increase regulatory scrutiny, burden AI companies, and encourage licensing agreements with content creators.
What has worked for OpenAI in lawsuits globally?
OpenAI notes that it has been sued for copyright infringement in 13 countries but none of those cases have resulted in injunctions.
For instance, the US District Court for the Southern District of New York recently dismissed Raw Story Media Inc.’s Digital Millennium Copyright Act claims against OpenAI. Raw Story had alleged that its copyrighted journalism was scraped and used to train ChatGPT, leading to potential plagiarism in responses.
The court ruled that Raw Story failed to show a substantial risk of ChatGPT producing verbatim copies of its work. The court noted that ChatGPT synthesizes vast amounts of information, making the likelihood of specific plagiarism remote, and dismissed claims of copyright infringement.
Also read | What if ChatGPT’s AI search engine clicks with users?
OpenAI’s primary defence in the ANI case is jurisdictional, arguing that its operations do not take place in India. The company is also likely to rely on “fair use" and “transformative use" arguments, contending that AI outputs are not exact reproductions of original works and often add new meaning or expression.
Sohini Mandal, founder of Nilaya Legal, explained: “Fair use and transformative use are central to OpenAI’s defence globally. Courts may conclude that the use is fair if it is sufficiently transformative and does not substitute the original work."
What about the New York Times’s case against OpenAI?
The most prominent case challenging OpenAI’s language models is the lawsuit filed by the New York Times in the US. The Times has accused OpenAI of using its copyrighted articles without authorization to train its models, alleging misappropriation of its content for profit.
A ruling is still pending in this case. But NYT’s lawsuit against OpenAI and other similar cases globally could shape the future of AI regulation and copyright law, setting a global precedent for how AI models are trained and how intellectual property is protected in the digital age.