When Copyright, AI, and Regulation Interact: Where Does Accessibility Stand?

A friendly blue robot holding a glowing light bulb stands beside the text “When Copyright, AI, and Regulation Interact” written in a bold, stencil-style font on a black background. Featured image for article: When Copyright, AI, and Regulation Interact: Where Does Accessibility Stand?

Summary

This article examines the intersection of copyright, artificial intelligence, and accessibility rights in India. It explains how the legal framework—comprising the Rights of Persons with Disabilities Act, 2016 and Section 52(1)(zb) of the Copyright Act—grants persons with disabilities the right and exception to access content in accessible formats. While these provisions are clear, the actual availability of accessible content remains minimal. The article shows how AI tools have recently made it feasible for persons with disabilities to independently convert content into accessible formats, overcoming historical barriers of cost, time, and dependency. However, ongoing copyright litigations, platform-level restrictions, and incoming AI regulations are now creating new barriers. These shifts risk undermining both legal rights and the practical ability to access content, unless accessibility is factored into future legal and technological decisions.

Introduction

In India, accessibility is no longer just a policy goal or aspirational concept. It is now both a fundamental right and a statutory right. The Rights of Persons with Disabilities (RPwD) Act, 2016 mandates public and private establishments to make information, content, communication systems, and digital services accessible to persons with disabilities. Courts have also recognised accessibility as a fundamental right of persons with disabilities in different contexts ranging from education and employment to leisure, sports, and transport.

In the context of information, content, and knowledge, accessibility-related provisions have also been codified under the Copyright Act, 1957 (Copyright Act). Section 52(1)(zb) of the Copyright Act provides an exception that allows copyrighted works to be converted into accessible formats for the benefit of persons with disabilities. The objective of this exception is to prevent copyrights from becoming a barrier to access of works by persons with disabilities. The Marrakesh Treaty, to which India is a signatory, mandates this exception on all its member states.

The legal framework under the RPwD Act and the Copyright Act provides the foundation for accessibility of information, content, and books to persons with disabilities in India. But how much of a difference are the accessibility right and copyright exception making to persons with disabilities? The straightforward answer is: not much. For example, Bookshare, an online platform that provides accessible books, has only 1.3 million titles, which is a small percentage of an estimated total of about 160 million books published so far. In the field of law, the number of accessible books available is much less, and based on experience, only one out of a thousand books and materials are available in accessible formats.

In the short time that AI applications and technologies were fully functional, accessibility took a huge step forward. But this seems to be regressing, and unless appropriate steps are taken, the advantage of AI to persons with disabilities may once again be limited to a small percentage of the content being created.

From Legal Right to Practical Accessibility

To access content and information directly or through assistive technologies, persons with disabilities require the content to be made available in accessible formats. The process of converting content into an accessible format is more complex than it may appear. A scanned book, for instance, may be readable to some extent, but it is not accessible to most persons with visual or cognitive disabilities. Assistive technologies like screen readers can help, but they cannot work effectively unless the content is structured in a specific way.

To be accessible, a document must follow certain principles. Headings and paragraphs must be clearly defined. Tables and lists must be marked correctly. Images need to be described using appropriate alternative text. Graphs must carry contextual explanation. Videos must include audio descriptions and accurate captions. Without these elements, content may technically exist in digital form, but it will remain inaccessible to persons with disabilities even with assistive technologies.

Creating such structured and tagged content has traditionally been a labour-intensive activity. It takes skill, time, and financial resources. For example, converting a scanned textbook into an accessible format suitable for screen reading may take between forty to sixty hours, depending on length and complexity. Describing diagrams, graphs, or illustrations requires subject knowledge. And, adding captions or audio descriptions to a video demands both effort and technology. Because of these requirements, most persons with disabilities have historically depended on content owners or nonprofit organisations to create accessible format works. This has meant delays, inconsistencies, and limited availability. As indicated earlier, only a very small percentage of books are available in accessible formats, and as millions of books are being published every year, the gap is increasing despite the accessibility right and copyright exception.

Artificial Intelligence as an Accessible Format Enabler

In recent years, artificial intelligence technologies have altered this landscape. Several AI tools can now carry out content structuring, tagging, and formatting with a relatively high degree of accuracy. AI-powered systems can detect document structure, identify headings, extract paragraphs, tag images with contextually relevant descriptions, and even generate audio descriptions for videos. These tasks, which once took days or weeks to complete manually, can now be done in hours or minutes with minimal human intervention.

The impact of this shift is significant. For the first time, persons with disabilities now have technologies that allow them to convert copyrighted content into accessible formats independently, efficiently, and at low cost. This independence reduces reliance on copyright owners, non-profit organisations, or third parties and enables timely access to books, documents, educational materials, and online content.

These AI-driven tools do not just make consumption easier through text-to-speech or captions. They address the root challenge of accessibility by transforming content into a format that assistive technologies can work with. They allow individuals to move from a state of passive waiting for accessible format content to one of active engagement in creating such content.

The Legal Framework Supporting Conversion

Under the Indian Copyright Act, the accessibility exception to copyright infringement is broad and reasonably well-defined. Section 52(1)(zb) permits reproduction, adaptation, and communication of works in accessible formats by or on behalf of persons with disabilities. The exception is broad enough to cover conversion into any format that meets the specific needs of persons with different types of disabilities, and any person can carry out conversion and share or communicate accessible formats to persons with disabilities.

Though this exception has been in place for 13 years now, it was of limited practical use in the absence of affordable and efficient tools to convert works into accessible formats. AI technologies have filled that gap. They have made an otherwise cost, time, and effort-intensive task easy, and have given persons with disabilities the power to convert works into accessible formats without depending on others.

Accessibility Restrictions Emerging Through Platform Practices

For a couple of years, AI tools did not restrict or control conversion of works into accessible formats, and persons with disabilities could carry out this activity without much difficulty. But now, new restrictions are starting to emerge with respect to copyrighted works. Many AI platforms have started placing limitations on how users can interact with copyrighted content.

Several widely used platforms, including generative AI tools such as ChatGPT and Gemini, now prevent users from uploading or converting entire books or documents. Their terms of use expressly prohibit full document processing if the content is copyrighted. Some platforms even automatically block access or impose use limitations when copyrighted material is detected. If users insist that the action is for accessibility purposes, they are often directed to use alternative applications specifically designed for accessibility purposes.

These changes are not necessarily the result of new legal provisions. They are the outcome of ongoing copyright litigations and risk management decisions taken by technology companies. In trying to avoid infringement claims, these companies are choosing to limit functionality, even if it means reducing the usability of their tools for persons with disabilities.

This shift creates a new kind of access barrier. It moves the restriction from law to technology. Even though the law permits conversion into accessible formats, the tools available to users may no longer support it. This can lead to a situation where a legal right cannot be fully exercised because the supporting technology no longer facilitates accessible content creation.

Accessibility and Regulatory Developments in Artificial Intelligence

The situation is further complicated by emerging regulatory discussions around artificial intelligence. In India and elsewhere, there are growing efforts to regulate AI systems, particularly in sensitive sectors such as healthcare, education, employment, and financial services. These regulations often include requirements for security, transparency, and risk management. Some frameworks propose limiting the processing or output of AI tools in certain contexts.

If such regulatory controls are applied broadly, they may affect the use of AI tools for accessibility purposes. Restrictions on data processing, document conversion, or model behaviour could make it more difficult to convert content into accessible formats. These controls may not be aimed at accessibility, but they can have unintended effects unless accessibility concerns are specifically addressed during the regulatory design process.

While it may seem so, this risk is not fully theoretical. Earlier experiences with anti-circumvention laws in different countries have shown how legal and technical protection measures can block accessibility, even when the underlying use is lawful. Unless exemptions or safeguards are built into new regulations, similar outcomes can arise in the AI context as well.

Keeping Accessibility in Mind

There is a need to make sure that the platforms and technologies used to exercise the accessibility right and copyright exception by way of conversion to accessible formats remain usable and legally permitted.

Technology providers, content owners, and regulators must take into account the accessibility dimension while designing platform restrictions or legal controls. While protecting copyright interests and regulating AI use are legitimate goals, these measures should not have the effect of limiting accessibility for persons with disabilities.

It is equally important to recognise the shift that AI has enabled. What was once a specialist task involving high cost and effort is now increasingly within the reach of individuals. The value of this change must be preserved. Any future regulation or enforcement strategy should be evaluated not only from the standpoint of copyright compliance or AI safety, but also from the perspective of whether it supports or obstructs accessibility.

Conclusion

Accessibility is today mandated under international treaties and conventions, and domestic laws. With the arrival of artificial intelligence tools, the practical barriers to exercising these rights have started to come down. AI has enabled persons with disabilities to create accessible content independently, making the copyright exception meaningful in day-to-day life. However, application, technology and platform level restrictions and incoming AI regulations have begun to create new limitations. These developments may not directly target accessibility, but their effects are being felt by persons with disabilities who rely on AI tools to convert content into accessible formats.

As the legal and technological landscape continues to evolve, it is important to make sure that accessibility does not become an unintended casualty. The right exists, the tools are available, and the law supports  accessibility, it is now essential that future changes continue to support, rather than restrict, the ability of persons with disabilities to access content in the formats they need.

References

Author: Dr. Kalyan Kankanala

Dr. Kalyan Kankanala is a practicing intellectual property (IP) attorney and author. He is a senior partner at BananaIP Counsels, a well-known IP firm based in Bangalore, India. His writings cover a range of topics relating to IP law, business, and policy, and he has authored several books and articles in the field. He has been contributing to this blog since 2007. The views expressed here are his own and do not represent those of BananaIP Counsels or its members.

Category