The India-AI Impact Summit 2026 is on in New Delhi and I am reminded of a story. A former colleague, Nirmita, who is visually impaired, once found herself in an absurd legal position. She could not legally purchase a book from the United States in a disability-friendly format called DAISY (Digital Accessible Information System), even though I, as a sighted reader, was able to purchase any print or e-books I wished. This was because of the vagaries of copyright law.

To address this issue, our non-governmental organisation, together with international coalitions of disability rights organisations, engaged in years of advocacy at the international level. These efforts ultimately led to the creation of the Marrakesh Treaty, which enables the cross-border exchange of accessible-format books as well as national exceptions for visually impaired persons to use technology to convert books into accessible formats when publishers do not make them available. The copyright industry -- from book publishers to the movie industry -- strongly opposed the treaty, which sought to establish a 'right to read' for visually impaired persons as any exception to copyright law was viewed as fundamentally unacceptable, even if it was at the expense of denying access to the visually impaired.

It is now copyright maximalism

The struggle of visually impaired persons against overly rigid copyright laws highlights a fundamental problem: copyright has expanded far beyond its original purpose, and copyright maximalism now actively obstructs the creation of and access to knowledge. This debate has taken on a renewed vigour thanks to Artificial Intelligence (AI) models, many of which turn out to be useful only when they have large quantities of training data (which, for language models, inevitably means copyrighted works). But before looking at AI, copyright needs to be understood in a historical perspective.

We have had art for far longer than we have had copyright. The Statute of Anne, widely seen as the first copyright law, was passed in Britain in 1710, after the era of Shakespeare and Milton. The British brought copyright law to India in 1847. The current Copyright Act is from 1957. In 1710, the law granted authors a limited monopoly of 14 years, with the possibility of one renewal. The monopoly right would only vest if it was specifically registered and multiple copies of the book deposited for distribution among libraries and universities.

Under the current law, the monopoly right goes well beyond the act of publishing, vests automatically the moment 'a work' is created, and lasts for the author's entire lifetime plus 70 years posthumously. So, the thousands of random Instagram posts and notebook doodles that you have made are all protected under copyright law for centuries. While the public domain was once the default, now a nearly-perpetual copyright monopoly is the default, regardless of the commercial potential of the work or ambitions of the creator. This fundamental change in the nature of the law has deleterious consequences.

Findings from a study

As part of a research project by LIRNEasia, a Sri Lankan think tank, we studied the data governance regimes of seven countries in South and Southeast Asia. With respect to copyright, we found that in four out of seven, the law made web search engines and AI training illegal. Web search engines need to copy as much of the Web as they can (a process called 'crawling'), effectively creating a mirror copy of all that is reachable on the web through links, but permissionless copying is prohibited by copyright law. Except in the Philippines and Sri Lanka (which have a flexible 'fair use' exception) and India (which, in 2012, introduced a specific exception for 'the transient or incidental storage' for 'providing electronic links, access or integration'), no other country in our study provided an exception, meaning AI training is effectively illegal in most countries we surveyed.

This does not make sense. Web search engines and AI models do not view copyrighted materials as scribblings or poems or art the way humans do; for programs, it is merely 'data' for statistical purposes. Recognising this, many jurisdictions such as the European Union, Japan and Singapore have adopted 'text and data mining' exceptions in their copyright regimes, while others such as Hong Kong and South Korea are in the process of doing so. Japan's law allows for an exemption for "Exploitations not for enjoying the ideas or emotions expressed in a work" (i.e., use by machines), and permits "using the work in data analysis". This is sensible: copyright was never meant to cover mechanistic uses.

By not allowing for a broad text and data-mining exception, India has created a pall of legal uncertainty over the collection of the training data for many forms of AI. And by not having a flexible, general and open-ended exception (as countries such as Singapore and the United States do), India ensures that copyright law will always hamper technological developments.

There are separate concerns around the outputs of generative AI substituting creative labour. But copyright is meant to be about encouraging creativity, not about protecting jobs. Further, copyright law has never prohibited learning from examples and imitating -- every artist studies predecessors, and every writer reads widely. Technology has always displaced jobs -- we have far fewer rickshaw pullers, telegraphists, pankhaawallahs, stenographers, lift operators, bank tellers, typesetters, darkroom technicians and draughtsmen -- yet, it has created new jobs as well. The advent of photography reduced the demand for portraitists, but enabled new forms of creativity and access to knowledge. We do not know what the impact of generative AI will be: we might, in the future, need greater government grants for arts and culture, or to strengthen the cooperative movement, potentially funded by taxes from large AI companies. But these ought not be dealt with in copyright law.

Creativity, access should be promoted

What copyright law should protect, however, are contributions to the commons. Open-licensed AI models and datasets exemplify this -- developers and researchers absorbing massive computational costs to create what enables others to be creative. These models add to the common heritage of mankind rather than subtracting from it. Copyright law should encourage such contributions, not hinder them with the same restrictions designed to prevent commercial exploitation. Governments are also uniquely positioned to curate high-quality locally-relevant datasets for public benefit, they should establish safe harbour provisions that protect such datasets from copyright claims, at least when used for training open-source models.

We have seen copyright law repeatedly being weaponised to block beneficial technologies under the guise of protecting creators. The Authors Guild in the U.S. used copyright to block Amazon Kindle's "Read Aloud" function, despite it being assistive technology that enabled visually impaired persons to listen to books they had legally purchased. Current copyright law blocks technologies that could democratise access to knowledge, unleash creativity, and drive innovation -- the very things that copyright was meant to foster. India's hosting of the AI Summit is the moment to act: it must lead efforts for all nations to adopt flexible exceptions that serve creators and the public, rather than the copyright industry. We need to bring copyright law into the 21st century by returning to its roots.

Pranesh Prakash is a tech law and policy consultant, working with think tanks, tech companies and universities. This article is based on research funded by LIRNEasia, based on a grant from IDRC, a Canadian taxpayer-funded research donor. The views expressed are personal

India's moment to restoring balance to copyright