ARTICLE
20 January 2025

Legal Perspectives On AI Governance In India: A Summary Of The AI Governance Guidelines Report

Artificial Intelligence ("AI") holds transformative potential for India's economy and society. However, this power also brings legal, ethical, and regulatory challenges that necessitate a comprehensive governance framework.
India Technology

Artificial Intelligence ("AI") holds transformative potential for India's economy and society. However, this power also brings legal, ethical, and regulatory challenges that necessitate a comprehensive governance framework. The "Report on AI Governance Guidelines Development" ("Report") outlines India's roadmap for addressing these challenges through a legally sound and inclusive approach.

  1. The IndiaAI Mission: A Foundation for Governance

On March 7, 2024, the Government of India approved the IndiaAI Mission with a budgetary allocation of Rs.10,371.92 Crore. This initiative aims to establish a robust AI ecosystem built on seven key pillars, including Safe & Trusted AI, which focuses on ethical AI development and deployment.

The Safe & Trusted AI pillar emphasizes creating indigenous tools for bias mitigation, algorithm auditing, and ethical certifications. These initiatives align with global frameworks like the Organisation for Economic Co-operation and Development of Artificial Intelligence Principles, adapted to India's unique socio-economic and legal landscape.

  1. Legal Principles for AI Governance

The Report highlights the need for a legally grounded framework based on principles such as:

  • Transparency: AI systems must disclose their capabilities, limitations, and decision-making processes. Users should know when they are dealing with AI systems.
  • Accountability: Developers and deployers must bear responsibility for outcomes, with liability frameworks aligned with existing laws.
  • Safety, Reliability and Robustness: Regular audits and monitoring of AI systems should be mandated to mitigate risks and ensure compliance in accordance with their specifications.
  • Privacy and Security: AI systems should be developed, deployed and used in compliance with applicable data protection laws and in ways that respects users' privacy.
  • Fairness and Non-Discrimination: AI systems should be fair and inclusive to and for all. They must not discriminate or create biases or prejudices against or preferences in favour of individuals, communities or groups.
  • Human Oversight: AI systems must remain subject to human intervention, judgment and oversight. Mechanisms should be in place to respect the rule of law and prevent adverse outcomes on society.
  • Inclusive and sustainable innovation: AI systems should be used to pursue beneficial outcomes for all and to deliver on sustainable development goals.
  • Digital by design governance: AI governance should leverage digital technologies to redesign systems and processes, adopting necessary techno-legal measures to operationalize principles and ensure compliance with applicable laws.

The above principles provide the legal backbone for AI governance and align with India's constitutional guarantees of equality, privacy, and justice.

  1. Key Legal Challenges and Gaps

a) Enforcing Existing Laws

  • Deepfakes and Malicious Content: Current laws like the Information Technology Act, 2000 ("IT Act"), Indian Penal Code, 1860 (now Bhartiya Nyaya (Second) Sanhita, 2023), and sectoral regulations address cybercrimes, identity theft, and misuse of AI for creating malicious synthetic media. However, technical tools like traceability and watermarking are needed to ensure compliance and effective enforcement.
  • Cybersecurity: Laws such as the IT Act and Digital Personal Data Protection Act, 2023 regulate cybersecurity and data protection. Sector-specific guidelines from sectoral regulators like the Reserve Bank of India, Securities and Exchange Board of India and Insurance Regulatory and Development Authority exist but require updates to address AI-specific risks.
  • Intellectual Property Rights: Training AI on copyrighted data without permission is likely an infringement under Indian copyright law. Clarity is needed on issues like liability for infringing AI outputs and copyrightability of AI-generated works. Techno-legal tools for tracing copyrighted data usage can aid compliance.
  • Bias and Discrimination: AI systems risk reinforcing biases, making it difficult to detect or prove intent. Existing laws must address such biases, with transparency and risk mitigation mechanisms for deployers of AI.

b) Ensuring Transparency and Responsibility

A transparent ecosystem is vital for traceability of data, models, and actors. Regulators need clear insights into AI systems' lifecycle and liability allocation. Both sectoral and cross-cutting governance frameworks are necessary to address risks comprehensively.

c) Adopting a Whole-of-Government Approach

Fragmented oversight by different regulators risks inefficiencies and gaps. A coordinated approach is required to address cross-sectoral issues and ensure a unified governance roadmap for AI development and deployment.

  1. Recommendations for Legal Reforms

a) Inter-Ministerial AI Coordination

The Report advocates establishing an Inter-Ministerial AI Coordination Committee to harmonize sectoral laws and streamline AI governance. This body would ensure legal clarity across domains such as finance, healthcare, and consumer protection.

b) Technical Secretariat

A technical secretariat is recommended to oversee risk assessments, map stakeholders, and establish technical standards ("Technical Secretariat"). It would also facilitate the development of sector-specific datasets, crucial for evaluating AI fairness.

c) AI Incident Database

An AI incident database is proposed to document risks and inform legal and policy interventions. This database would function as a non-punitive mechanism to encourage reporting and foster a culture of accountability.

d) Transparency & Governance

The Technical Secretariat should promote voluntary industry commitments, including transparency reports, risk assessments, and adherence to AI principles. These commitments, tailored by sector, complement legal frameworks, encourage self-regulation, and minimize prescriptive regulations.

e) Technological Measures

The Secretariat should explore tools like watermarking and labelling to address AI risks and ensure compliance, content provenance, and public awareness.

f) Legal Integration with the Digital India Act (DIA)

The proposed Digital India Act should harmonize existing AI-related laws, strengthen grievance redressal systems, and establish adjudicatory mechanisms tailored to digital industries.

  1. Future Directions for AI Governance

The Report identifies unresolved legal questions that warrant further exploration:

  • Training on Copyrighted Data: Should AI systems be allowed to train on copyrighted datasets without individual permissions?
  • Market Dominance: How can antitrust laws address the concentration of computational resources and data among a few entities?
  • Definition of AI: Should AI be defined narrowly or broadly within the legal frameworks?
  • Safe Harbor Protections: Can AI deployers claim immunity under intermediary liability provisions like Section 79 of the IT Act?

These issues highlight the need for dynamic legal frameworks that balance innovation with public interest.

  1. Conclusion

The Report provides a comprehensive legal roadmap for AI governance in India. By integrating ethical principles, addressing regulatory gaps, and fostering industry collaboration, India aims to lead in responsible AI innovation. The proposed measures prioritize inclusivity, transparency, and accountability, ensuring AI aligns with constitutional values and legal standards. As AI technologies evolve, this framework offers the flexibility to adapt while safeguarding the rights and interests of all stakeholders.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.

Mondaq uses cookies on this website. By using our website you agree to our use of cookies as set out in our Privacy Policy.

Learn More