India Will Publish an AI Regulation Paper Within a Month

India's position on AI regulation has been clear for almost a year: use the laws already on the books, ask industry for voluntary commitments, and only write new rules if that fails. On Thursday that position moved.

Speaking at the India launch of the World Bank's World Development Report 2026: The Promise of Artificial Intelligence, Electronics and IT Minister Ashwini Vaishnaw said MeitY should publish a consultation paper on regulating AI within a month. He described it as "probably the base for the regulation which has to come."

What the paper will cover

Vaishnaw named four elements the paper will rest on: safety, a "human-first" approach, skilling, and making sure no one is left behind.

Deepfakes get their own attention. He said their quality is improving every day and that people are being lured into believing and doing the wrong things because of them. Harms linked to psychological disorders will also be part of the paper.

He also said a law on its own won't do the job. India wants a "techno-legal" approach with stakeholders on board, and he said the bulk of the responsibility will have to be carried by industry, citing cybersecurity, safety and user harm.

Why this is a shift

The India AI Governance Guidelines released in November 2025 deliberately avoided proposing a dedicated AI law. They leaned on the IT Act, the Digital Personal Data Protection Act, consumer law and sectoral rules, and gave voluntary commitments 9 to 12 months before MeitY would consider mandatory requirements.

That window is running out. MeitY Secretary S. Krishnan said the World Bank report recommends the same sequence India followed (existing laws, then voluntary compliance, then regulation) and added that "maybe some of those times have come." He also pointed out that restricting the flow of AI models between countries is already regulation in all but name.

The compute side

There were also numbers on compute:

• MeitY issued an RFP on Wednesday for a five-year GPU supply agreement, with demand put at 10,000 GPUs and a first tender for 5,000.
• Contracts for 3,000 GPUs were issued last week.
• A term sheet signed during the Prime Minister's UAE visit adds the equivalent of roughly 10,000 GPUs.
• Government AI procurement will run through empanelment rather than a single tender, so empanelled firms and startups can bid for work from any department.

What enterprises should do now

A consultation paper isn't a law. It does set the vocabulary the law will use, so a few things are worth doing before it lands:

1. Find where you generate or distribute synthetic media. Marketing, customer service voice bots and internal training content all count. Deepfake obligations look likely to come first.
2. Write down your voluntary commitments, if you haven't. If the government is about to judge whether voluntary compliance worked, you want to be able to show what you committed to and how you measured it.
3. Plan to respond to the consultation. The minister said industry will carry most of the responsibility. Sector bodies in BFSI, healthcare and telecom should be ready to say what that responsibility looks like in practice.

The thing to watch over the next month is how much of the November 2025 guidelines survives. The human-first, skilling-heavy framing points to continuity. The specific language on deepfakes and user harm points to the first hard obligations.

Source: ThePrint — India steps up AI regulation push with consultation paper likely next month