News agency Asian News International’s November 19 copyright infringement suit against United States-based OpenAI in the Delhi High Court has finally brought before an Indian judge a complex legal question that courts across the world have been wrestling with: are artificial intelligence platforms liable for copyright infringement?
OpenAI operates the popular advanced artificial intelligence chatbot, ChatGPT which is trained on vast amounts of text from the internet.
Legal experts that Scroll spoke with expressed unanimity that this is an unresolved legal question, with no global precedent. How the Delhi High Court resolves it will have ramifications for how artificial intelligence platforms develop and operate in India.
The lawsuit
In its lawsuit against OpenAI, ANI claimed that ChatGPT uses its material to train its language model, storing and reproducing copyrighted content for commercial purposes without permission. OpenAI countered, stating it has no servers or offices in India and denied reproducing ANI’s material in the country.
OpenAI also claimed that since October, on the receipt of the lawsuit, it had blocked ANI’s website for the purpose of training ChatGPT.
ANI further alleged that ChatGPT generates responses using ANI’s content verbatim and falsely attributes statements, including fabricated ones, to the agency. These inaccuracies, known as so-called hallucinations in the field of artificial intelligence, threaten ANI’s reputation and risk spreading fake news.
The news agency has sought damages of Rs 2 crore, as well as directions to stop OpenAI from storing or using its work.
Citing the complexity of the case, the court has said it will appoint an amicus curiae, or friend of court, to assist the bench in the matter. It has scheduled the next hearing for January.
Tricky legal question
In India, disputes pertaining to copyright infringement are governed by the Copyright Act, 1957. The act protects the rights of creators over their original works. It ensures fair usage, prevents unauthorised reproduction and promotes creativity by safeguarding intellectual property.
Copyright violation occurs when someone uses, reproduces or distributes a copyrighted work without the owner’s permission or legal justification. This includes copying, sharing or profiting from the work unlawfully.
For training themselves, artificial intelligence platforms rely on text content across the internet. Due to lack of regulations so far, they have not been required to seek approval or licenses from any online content creators for using their content.
Because it is nearly 70 years old, the Copyright Act in not equipped to deal with the potential for copyright infringement posed by artificial intelligence platforms that have proliferated only in the last few years.
The Delhi High Court will have to thus grapple with novel questions of law. It will have to set the precedent on whether the process of training artificial intelligence models on data scrapped from the open internet involves copyright infringement.
“Indian law provides what is known as the fair dealing exception,” Aman Taneja, partner at Ikigai Law, a law and policy firm specialising in technology and emerging businesses, told Scroll. “Whether this exception can be relied on for training AI models will have to be evaluated by the court.”
The fair dealing exception under Indian copyright law permits limited use of copyrighted material without the owner’s consent for specific purposes, such as research, criticism, review, reporting or teaching. For instance, “copyright protects original expression, but not facts”, said Taneja. The objective of this exception is to ensure a balance between protecting creators’ rights and promoting public access to information and ideas.
Jameela Sahiba, Senior Programme Manager at tech policy think-tank The Dialogue, agreed. “The court’s interpretation of the terms ‘adapation’ and ‘reproduction’ in the Copyright Act is key in this case,” she told Scroll.
According to Taneja, “These are complex and nuanced questions.”
Unresolved across the world
This is the story across the world, as courts in several countries have been flooded with copyright infringement suits by media organisations against artificial intelligence platforms. “What amounts to copyright in such cases is globally contested,” said Sahiba.
OpenAI acknowledged in the Delhi High Court last week that it is facing 13 similar lawsuits in the United States, apart from two in Canada and one in Germany, ever since it began operating in 2022.
While all these cases are pending, in none of them has any court reached even a preliminary finding that OpenAI had broken copyright law.
Karthika Rajmohan, Associate Policy Counsel at Internet Freedom Foundation, a digital rights organisation, explained that the task before the courts is not easy. “This is an extremely tricky policy question,” he said. “No jurisprudence has been offered by courts in any other jurisdiction yet.”
Even once any of these cases are decided, “these will likely be appealed till the highest available courts, so it might be some time before we see the jurisprudence firming up”, Taneja said.
Crucial implications
Due to the unprecedented nature of the case, the Delhi High Court’s adjudication of the case will be pivotal, experts told Scroll. “The court’s decision will impact the way AI platforms train and develop their models in India,” Sahiba said.
India is the second-largest national market for ChatGPT after the US, with 11.1% of its 20 crore weekly active users from India. Rajmohan explained that “the court will have to perform a balancing act between protecting creativity and preserving access to knowledge”.
She laid out another crucial implication of the decision. “OpenAI has had unrestricted access to the internet thus far,” she said. “Such access may not be available to other entrants in the industry, depending on the court’s decision.” This could be due to the court expressly forbidding artificial intelligence platforms from accessing copyrighted content online. Alternatively, the opt-out resolution offered by OpenAI – wherein it voluntarily blocked ANI’s content for its training – may come to become the industry standard.
This might lead to a competition law issue, according to Rajmohan, because of OpenAI having benefited from untrammeled access to online content that will be denied to other players in the industry.
Scroll had reported earlier this year that the Union government’s belligerence regarding alleged bias in artificial intelligence platforms would hurt the development of the industry in India.
