India AI & Infrastructure 2026-09-20

India Chose Guidelines Over a Law -- a Deliberate Bet Against the US State-by-State Model

MeitY's AI Governance Guidelines are voluntary and principles-based, adapting existing law rather than creating a dedicated AI statute -- the opposite bet from the state-by-state hard-law patchwork building up in the US this year.

India's Ministry of Electronics and Information Technology unveiled AI Governance Guidelines under the IndiaAI Mission, and the notable choice is what they deliberately are not: not a new AI-specific law, not mandatory, not enforced through a dedicated regulator. The guidelines are explicitly voluntary and principles-based, leaning on existing law (data protection, consumer protection, sector-specific rules) rather than creating a horizontal AI statute the way the EU AI Act does.

This is a legible, deliberate strategic choice, not indecision -- India's stated rationale is prioritizing innovation and adoption speed while existing law absorbs the risk cases as they appear, rather than front-loading compliance cost before the market and the risks are fully understood. It's close to the opposite bet from what's building in the US this year, where -- as covered here yesterday -- five additional states passed AI laws with real enforcement teeth (Connecticut's SB 5 disclosure requirements, multiple states restricting AI in health-insurance decisions) even as federal guidance stays non-binding.

The infrastructure underneath the guidelines is further along than the policy framing might suggest: AIKosh, the national data/model repository under the IndiaAI Mission, already hosts more than 9,500 datasets and 273 sectoral models, and the FutureSkills workforce program has trained or is actively upskilling over 1 million people in AI-adjacent competencies. The governance approach and the infrastructure buildout (see the companion piece on the $260B pledge and 3.5GW under construction) are explicitly coordinated under one mandate, not separate initiatives running in parallel.

Whether "lightweight guidelines now, adapt as needed" beats "hard law state-by-state now" is a genuinely open empirical question this year is going to start answering -- India is betting that speed-to-adoption matters more than compliance certainty at this stage of the technology, while the US patchwork is betting that verticals like health and employment can't wait for a mature market to self-correct. Both are real policy positions, not the absence of one.

India's voluntary, principles-based guidelines and the US's emerging state-by-state hard-law patchwork are opposite bets on the same open question -- whether AI regulation should front-load compliance cost now or adapt existing law as specific harms appear -- and both are deliberate strategic choices worth watching for which one actually holds up as adoption scales.