Inter-ministerial committee needed to strengthen AI governance, says Centre’s IndiaAI Mission

In a recognition that artificial intelligence (AI) has impacts on a wide range of fields, a Union government panel mandated to develop AI guidelines has recommended that an inter-ministerial committee be set up to enforce AI rules and ensure effective governance of the country’s emerging AI ecosystem.

An advisory group of the Union Ministry of Electronics and Information Technology’s IndiaAI Mission headed by the Principal Scientific Advisor is seeking public feedback on a report released on Monday (January 7, 2025) by its AI guidelines sub-committee. The report proposes a “coordinated, whole-of-government approach to enforce compliance and ensure effective governance as India’s AI ecosystem evolves”.

The panel’s report is notable as it is the most detailed articulation yet of the government’s approach towards AI in India. It appears that the Centre seeks to maximise gains from investments in AI, while having light regulatory frameworks to act as guardrails against any potential harm.

Multi-disciplinary expertise

Apart from an Inter-Ministerial AI Coordination Committee or Governance Group to oversee AI in India, the panel recommended that the IT Ministry should house a “technical secretariat” to bring in officers on deputation from other departments and regulators. This would help to “pool together multi-disciplinary expertise”, map stakeholders in the AI ecosystem, and assess risks, along with other research, the panel recommends.

The report outlines the following principles for AI governance: transparency of AI systems with “meaningful information on their development” and capabilities; accountability from developers and deployers of AI systems; safety, reliability and robustness of AI systems by design; privacy and security of AI systems; fairness and non-discrimination; human-centred values and ‘do no harm’; inclusive and sustainable innovation to “distribute the benefits of innovation equitably”; and “digital by design” governance to “leverage digital technologies” to operationalise these principles.

Life cycle approach

In order to operationalise these principles, the report says, policymakers must use a life cycle approach. This means they must look at AI systems at different stages of their development, deployment, and diffusion, during which distinct risks can occur. Second, there should be an “ecosystem view” of AI actors: data principals (those whose data is used), providers of data, AI developers, deployers, and end users must all be considered for a “holistic” view of AI in practice.

A “command and control” governance strategy “may not be able to adequately … promote the growth and expansion of this space,” the panel said. Instead, the report proposes a tech-enabled digital governance system. “This could allow players to manage liabilities in the chain leading to lightweight, but gradually scalable, regulatory control,” it says. On issues like deepfakes and cybersecurity, the report repeats the government’s position on the subjects in the past, that existing laws provide a legal framework to deal with these issues.

‘Enable self-regulation’

On the need for “transparency and responsibility across the AI ecosystem,” the report warns that “categorisations of systems purely based on computational capacity or data parameters may not be effective,” and that there could be sector-specific risks when looking at any AI system such as in health and banking. A “whole of government” approach has been recommended in the report, as “there is an inefficiency and possibility of gaps due to a fragmented approach” (emphasis in report).

The Committee that the report recommends should “apply and strengthen existing laws to minimise risk of harm due to use of AI,” it says, along with harmonising efforts across ministries, “enabling demonstrable self-regulation” among the AI sector, identify gaps with “multi-regulatory support,” create a policy environment conducive to “beneficial use-cases,” and promote deployment of “responsible AI applications” in individual sectors.