º¬Ðß²ÝÉçÇø

India stands at a critical juncture in the artificial intelligence (AI) revolution, where rapid proliferation of technology intersects with an evolving regulatory landscape. AI is increasingly ubiquitous across industrial and consumer applications, India’s AI regulation remains nascent, balancing innovation with risk mitigation.

The Indian government has allocated about USD11.7 billion for AI leadership, with the Ministry of Electronics & Information Technology (MeitY) the executive agency leading policy development through specialised committees.

The India AI Mission serves as the central catalyst, driving innovation through strategic programmes and public-private partnerships to democratise computing (along with India Semiconductor Mission), enhance data quality, build indigenous capabilities, attract talent, and foster industry collaboration.

º¬Ðß²ÝÉçÇø regulators, in their strive to achieve ¡°AI sovereignty¡± to build indigenous capabilities and to address unique domestic challenges, seek to do so without over-reliance or adoption of foreign frameworks.

This article examines key positions under Indian law on certain facets of AI development, training and deployment.

Intellectual property

The Copyright Act, 1957 is key to AI training and output ownership, as training datasets will likely include protected works, and outputs could be considered ¡°derivative¡± works.

There is no judicial protection for reproducing copyrighted work in its entirety, or in full. Reproduction is an exclusive right of the copyright holder, and unauthorised reproduction for commercial use would constitute infringement.

When assessing reproduction, courts apply three tests on a case-by-case basis:

1. 1. Quantum and value of matter reproduced;
   2. Purpose of reproduction; and
   3. Likelihood of competition between the original and the reproduction.

Fair dealing is a key defence to copyright infringement. Under section 52(1)(a) of the Copyright Act, fair dealing of any work is not infringement for:

1. 1. Private or personal use, including research;
   2. Criticism or review of that/any other work; and
   3. Reporting of current events/current affairs.

However, what constitutes ¡°fair dealing¡± depends on each case. Judicial precedents in RG Anand v Delux Films & Ors and The Chancellor Masters and Scholars of the University of Oxford v Rameshwari Photocopy Services hold that transformative use is key to the idea-expression dichotomy, with a certain degree of reproduction permitted if the purpose constitutes ¡°fair dealing¡± or otherwise benefits from limited, specific exemptions under section 52.

The use of copyrighted materials to train AI would have to be demonstrated as ¡°transformative¡± to constitute fair dealing. However, unlike ¡°fair use¡± is dealt with in the US, ¡°fair dealing¡± in the Indian context is limited in scope and Indian courts are yet to explicitly extend this to AI training.

AI training based on the collection-tokenisation-training process ¨C along with the sheer volume of training data, and rapid advances in technology on how AI models are trained and deployed ¨C does not fit within the traditional framework of how copyright violations are assessed.

Delhi High Court is currently considering these issues in the ANI v Open AI dispute: namely, whether:

1. 1. storing copyrighted data constitutes infringement;
   2. output generated using training data will be derivative work and constitute infringement;
   3. OpenAI can claim the ¡°fair dealing¡± exemption; and
   4. India has jurisdiction when OpenAI¡¯s servers are overseas.

The court¡¯s ruling is eagerly awaited and will likely set the tone for the future of AI training and copyright claims.

Until courts take a definitive stance, a balance currently seems aspirational between the legitimate interests of the author and the owners of AI models, with the result plausibly being a commercial monetary settlement. India is not alone in this journey of seeking a balanced solution to this problem.

IT legislation

AI is typically trained on non-personal data. However, data scraping may result in the collection and processing of personal data. Further, once AI tools are deployed, new data (personal or otherwise) from the user is collected and processed to provide tailored outputs.

Personal data collection and use are governed by the Information Technology Act, 2000 (IT Act); the IT (Reasonable Security Practices and Procedures and Sensitive Personal Data or Information) Rules, 2011 (SPDI Rules); and the recent but not currently enforced Digital Personal Data Protection Act, 2023 (DPDPA).

The IT ACT and SPDI Rules require that express consent is obtained to collect, process, disclose or transfer sensitive personal data and information. The DPDPA mandates that personal data (no longer limited to ¡°sensitive¡± information) be collected and processed only with the data principal¡¯s free, specific and informed consent.

Data principals must be notified of their personal data being collected, the purpose, and their rights to access, correct, update and erase the data, or withdraw consent to use it.

Section 17(2)(b) of the DPDPA does exempt processing of personal data for ¡°research, archiving or statistical pur-poses¡± provided it is not used ¡°to take any decision specific to a data principal¡±, and such processing is carried out in accordance with standards as may be prescribed.

While AI training could theoretically qualify as ¡°research¡± or ¡°statistical purposes¡±, the final determination will depend on how the government prescribes the standards, and whether AI training meets the specific conditions, particularly the requirement that no decision is made specific to individual data principals.

The IT (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, impose due diligence obligations on intermediaries (including AI companies), to prevent hosting of infringing, obscene or impersonating content.

While AI companies may seek protection under the safe harbour provisions of the IT Act, blanket protection is unlikely, given the dangerous ability of AI systems to generate deepfakes and spread misinformation.

The MeitY 2024 advisory imposed requirements for bias restriction and labelling of AI-generated output¡¯s fallibility, but implementation remains unclear following the advisory¡¯s withdrawal and absence of a clear standard.

Consumer protection

AI tools will likely fall within the definition of ¡°services¡± under the Consumer Protection Act, 2019.

The product liability regime could be used to hold AI product/service providers liable for harm to consumers, including due to faulty/biased algorithms, inadequate safety protocols, or unsafe software that divulges sensitive personal information. Liability is likely to be stringent in direct, high-risk use cases.

Sectoral regulations

As central AI legislation develops in India, sectoral regulators have issued targeted circulars and guidelines mandating disclosures specific to their respective markets and concerns.

1. 1. Securities and Exchange Board of India (SEBI)

• • 1. Intermediaries (circular 4 January 2019): reporting requirements for offering or using AI applications and systems.
    2. All entities in the mutual fund ecosystem (circular 9 May 2019): reporting requirements for offering/using AI applications and systems.
    3. Mutual Funds (circular 27 June 2024): all mutual funds using AI systems to report usage on a quarterly basis, to ensure full disclosure.
    4. Investment Advisers (regulations 16 December 2024; guidelines 8 January 2025): must disclose use of AI in operations, irrespective of scale and extent.
    5. Research Analysts (guidelines 8 January 2025; regulations 16 December 2024): must disclose use of AI tools, irrespective of scale and scenario, and be solely responsible for security, confidentiality, integrity of client data.
    6. Intermediaries (regulations 10 February 2025): anyone using AI tools, irrespective of scale and scenario, will be solely responsible for the privacy, security and integrity of stakeholders¡¯ data, the output arising from them, and compliance with applicable laws.

1. 2. Reserve Bank of India
      1. An August 2025 report on developing a Framework for Responsible and Ethical Enablement of AI in the financial sector (FREE-AI) urges lawmakers to legislate
         in a manner that balances innovation and risk. It prescribes seven guiding ¡°sutras¡± for AI adoption: trust is the foundation; people first; innovation over restraint; fairness and equity; accountability; understandable by design; and safety, resilience and sustainability.
      2. The report makes 26 recommendations under six strategic pillars: infrastructure, capacity, policy, governance, protection and assurance. It further recommends establishment of shared infrastructure by regulated entities to democratise access to data and computing, along with creation of an Al Innovation Sandbox.
   3. Department of Telecoms (DoT)

Based on extensive stakeholder consultations and expert input, the regulator unveiled a New Standard for Fairness Assessment and Rating of Artificial Intelligence Systems in 2023, outlining procedures for assessing and rating AI systems for fairness.

Takeaway

While AI transcends national boundaries, India’s pursuit of AI sovereignty¡ªtailored to serve its distinctive socio-economic landscape¡ªdemands a regulatory framework that is robust in oversight and aligned with evolving global standards.

As the world¡¯s fifth-largest economy, India is intensely conscious that risks posed by AI advancements may exacerbate inequalities and widen the digital divide without effective regulation.

AZB & PARTNERS
Villa B8, Withy Pool Villa Complex,
Boat Club Road, R A Puram,
Chennai 600028, India.
Tel: +91 44 2434 0145
Email: chennai@azbpartners.com