The ongoing AI summit in New Delhi offers India a historic opportunity to rethink copyright law in light of technological change.
What is the historical evolution of copyright?
The Statute of Anne (1710) – It was widely regarded as the first modern copyright statute, granted authors a limited monopoly of 14 years, renewable once.
Protection required registration and deposit of copies in libraries, reflecting a balance between author incentives and public access.
Indian Scenario – In colonial India, copyright law was introduced in 1847.
Post-independence, the Copyright Act, 1957 became the governing statute.
Over time, amendments have expanded the scope and duration of rights.
Present status – Today, copyright protection vests automatically upon creation and extends for the author’s lifetime plus 70 years.
The shift from a conditional, time-bound privilege to an automatic and prolonged monopoly has fundamentally altered the nature of copyright.
Earlier, the public domain was the default and copyright the exception. Today, virtually every original expression — including social media posts and unpublished notes — is protected for decades.
This expansion reflects what scholars term “copyright maximalism,” where protection is treated as absolute rather than balanced.
What is the issue between AI and the Centrality of Data?
Basis of AI – AI models, particularly large language models, rely on vast quantities of training data.
Web search engines and AI systems function by copying and indexing content from the internet through processes such as web crawling and text and data mining (TDM).
Technically, such copying may constitute infringement unless covered by exceptions.
Indian intervention – India’s 2012 amendment introduced an exception for “transient or incidental storage” for providing electronic links or integration.
While this offers limited protection for search engines, it does not clearly address large-scale AI training.
The absence of a broad, open-ended exception creates legal uncertainty for researchers and startups.
Interventions of Other countries – In contrast, several jurisdictions have adopted explicit TDM exceptions:
The European Union – It provides statutory text and data mining exceptions under its copyright directive.
Japan – It permits uses that are not for “enjoying the ideas or emotions expressed” in a work, thereby covering data analysis by machines.
Singapore – It has adopted flexible copyright exceptions conducive to innovation.
The United States – IT relies on an open-ended “fair use” doctrine, which courts have interpreted to allow transformative uses, including certain forms of data mining.
India – It is lacking a comprehensive TDM exception or a general fair use clause, risks falling behind in the global AI race.
Mechanistic use vs expressive consumption –A key distinction must be drawn between human consumption of creative works and machine processing of data.
AI systems do not “enjoy” or “experience” creative expression; they analyse patterns statistically.
Copyright law was originally intended to regulate expressive consumption and commercial exploitation, not mechanistic data analysis.
Japan’s legislative language explicitly acknowledges this distinction.
Such clarity ensures that copyright does not extend beyond its purpose.
If machines are merely extracting statistical correlations, treating such activity as equivalent to human reading stretches copyright beyond reasonable limits.
Copyright and Employment – Concerns about generative AI displacing creative labour are legitimate. However, copyright law is not a labour protection statute. Its objective is to incentivise creativity by granting limited monopolies, not to guarantee employment.
History demonstrates that technological change often displaces certain professions while creating new ones.
The advent of photography reduced demand for portrait painters but enabled new artistic forms.
Similarly, automation reduced the need for telegraphists and typesetters while generating new sectors.
If AI disrupts creative industries, policy responses should include public funding for arts, social security measures, or taxation of large AI firms.
Expanding copyright protection to block AI training would be an inappropriate and ineffective tool.
Protecting the Commons and Open Innovation – Copyright reform must also recognise the value of open-licensed AI models and datasets.
Developers and researchers invest substantial resources to create open-source systems that benefit society at large.
These contributions enrich the digital commons.
Governments can play a constructive role by curating high-quality, locally relevant public datasets and establishing safe harbour provisions to protect such datasets from copyright claims when used for non-commercial or open-source AI training.
This would align with the constitutional commitment to promote scientific temper and innovation.
Lessons from the Accessibility Debate – The struggle leading to the Marrakesh Treaty demonstrates how copyright law can be weaponised to block socially beneficial technologies.
For instance, the Authors Guild opposed the “Read Aloud” function of Amazon’s Kindle, despite its value as an assistive feature for visually impaired users.
Such episodes reveal a pattern: when new technologies emerge, copyright is often invoked to resist change.
Over time, however, balanced reforms enable both protection and progress.
What are the policy recommendations for India?
Introduce a Broad Text and Data Mining Exception – Explicitly permit copying for AI training and data analysis, especially for non-expressive uses.
Adopt a Flexible Fair Use Clause – Move towards an open-ended exception capable of accommodating future technologies.
Strengthen Public Domain and Commons – Incentivise open licensing and protect publicly funded datasets.
Ensure Accessibility Safeguards – Build upon the spirit of the Marrakesh Treaty to ensure technology enhances inclusion.
Separate Labour Concerns from Copyright Law – Address employment disruptions through economic policy rather than restrictive IP regimes.
What lies ahead?
Copyright law was conceived as a limited instrument to promote creativity and the dissemination of knowledge.
Over centuries, it has expanded into a near-perpetual monopoly covering almost all forms of expression.
In the age of AI, this expansion risks stifling the very innovation it was meant to encourage.
India stands at a crossroads. By embracing balanced exceptions and promoting the commons, it can demonstrate that intellectual property law need not be an obstacle to technological progress.
Instead, it can be recalibrated to serve both creators and the public interest.
The moment calls not for dismantling copyright, but for restoring it to its original purpose: fostering creativity, expanding access, and advancing human knowledge in the 21st century.
