MeitY unveils AI Governance Guidelines; outlines 'balanced' framework built on existing laws

Krishnan said the framework reflects the government’s view that technological development must remain centred around citizens and social impact.

“At the heart of it all is human centricity, ensuring AI serves humanity and benefits people’s lives while addressing potential harms,” he said.

The guidelines emphasise human-centric design, fairness, transparency, safety, and proportional accountability as key principles.

The government said the framework has evolved through a “whole-of-government and multi-stakeholder approach” involving extensive consultations with industry, academia, and civil society. Over 2,500 comments were received during the public consultation stage, according to MeitY.

Institutional design and governance mechanisms

The guidelines propose a layered governance structure spanning infrastructure, policy and regulation, accountability, risk mitigation, and capacity building. MeitY said the framework also envisages new institutional mechanisms to coordinate AI oversight and policy coherence. These include an AI Governance Group (AIGG) for cross-ministerial coordination, a Technology & Policy Expert Committee (TPEC) for technical and policy inputs, an AI Safety Institute (AISI) for standard setting and risk assessment, and sectoral regulators for domain-specific implementation.

The ministry said it would adopt a “phased approach” to implementation. In the near term, it will focus on establishing governance bodies, developing a risk classification framework, and setting up awareness and capacity-building initiatives.

Over the medium term, MeitY plans to operationalise an AI Incident Reporting System, launch regulatory sandboxes, and integrate AI governance with India's Digital Public Infrastructure. Longer-term plans include drafting sectoral rules and refining accountability standards as institutional capacity strengthens.

Accountability and risk mitigation

The guidelines call for graded responsibility across the AI value chain, encouraging transparency reports, grievance redressal mechanisms, and compliance-by-design models. They also propose voluntary “red-teaming” and human oversight for high-risk applications, alongside an India-specific risk taxonomy.

While the document introduces clearer institutional ownership of AI safety, it continues to rely on voluntary self-regulation and coordination across existing regulators.

Earlier advisory reports from MeitY had proposed an AI incident database and technical secretariat; the final framework incorporates these ideas but stops short of making disclosure or reporting mandatory.

Linking governance with application

The guidelines are part of the broader IndiaAI Mission, which aims to promote AI adoption across key sectors such as agriculture, healthcare, manufacturing, and natural resources.

“Our focus is primarily on innovation,” Krishnan said. “We will continue to take necessary action to protect citizens while ensuring India harnesses AI for inclusive and sustainable growth.”

The road ahead

The AI Governance Guidelines mark India’s most comprehensive articulation of its AI policy stance to date. However, much of the framework remains principle-driven, and the absence of binding enforcement mechanisms means its effectiveness will depend on how quickly MeitY and other institutions can translate it into operational rules.

By building on existing regulations and avoiding an immediate legislative approach, India has positioned itself between the European Union’s prescriptive AI Act and the United States’ market-led model. Whether this middle path succeeds, experts suggest, will depend on how effectively the proposed institutions — particularly the AI Safety Institute — are able to coordinate oversight, standard-setting, and transparency across sectors.