India’s MeitY Releases Draft AI Accountability Framework Mandating Algorithmic Audits for Platforms Serving Over 5 Million Users

Every major platform operating in India now has a countdown clock on its wall.

India’s Ministry of Electronics and Information Technology (MeitY) has published a draft AI accountability framework that would require any AI-driven platform with more than 5 million domestic users to undergo mandatory third-party algorithmic audits every 18 months. Non-compliance carries fines of up to ₹250 crore. The proposal opens a 60-day public comment window, signaling that India is moving from policy conversation to enforceable regulation — and doing so at a scale that will reverberate across global compliance teams.

—

What the Draft Framework Proposes

At its core, the draft targets recommendation engines and automated decision systems that shape what Indian users see, read, buy, and believe online. Platforms meeting the 5-million-user threshold must commission independent third-party algorithmic audits at 18-month intervals and submit findings to a designated regulatory body under MeitY’s oversight.

The audits are not merely procedural checkboxes. According to the draft, auditors must assess whether recommendation systems amplify harmful content, introduce discriminatory bias, or manipulate user behavior in ways that undermine informed consent. Platforms will also be required to maintain detailed technical documentation of their AI systems — including training data provenance, model update logs, and impact assessments — and make summaries of audit results publicly available.

The draft further distinguishes between “high-risk” AI applications, such as those influencing financial decisions or political content distribution, and lower-risk systems, with proportionally stricter audit requirements applied to the former.

—

Who Falls Under the Scope

The 5-million-user threshold is deliberately calibrated to capture the platforms that matter most to India’s digital public sphere. Meta’s Facebook and Instagram, YouTube, and homegrown platforms such as ShareChat and Moj fall squarely within scope, as do major e-commerce recommendation engines and algorithmically curated news aggregators.

For global platforms, this is not an abstract compliance exercise. India represents one of the largest user bases in the world for several of these services. A mandatory algorithmic audit regime effectively forces these companies to document and defend the mechanics of systems they have historically treated as proprietary black boxes. Compliance teams will need to build audit-ready infrastructure — not just for India, but potentially as a template for similar regulations emerging elsewhere.

Domestic platforms face a different kind of pressure. Startups and mid-size Indian tech companies operating recommendation-driven products will need to invest in audit infrastructure and legal compliance capacity that many have not yet developed. The draft does not currently propose a tiered structure for smaller companies that nonetheless cross the user threshold — a gap that industry groups are expected to flag during the comment period.

—

Why This Framework, Why Now

India’s move toward formal AI accountability did not emerge in a vacuum. The country has spent several years developing its broader digital governance architecture, including the Digital Personal Data Protection Act, and has watched the European Union’s AI Act progress from draft to enforcement. MeitY’s proposal reflects a clear strategic choice: India will not wait for international consensus before setting its own standards.

The emphasis on algorithmic audits as the primary enforcement mechanism is also significant. Rather than attempting to regulate AI outputs directly — a technically complex and politically fraught approach — the framework places accountability on the process by which AI systems are built, updated, and deployed. This aligns India with a growing global consensus that process-based regulation is more durable than outcome-based prohibition.

Digital rights advocates have long argued that recommendation engines operating in India have contributed to the spread of misinformation and, in documented cases, content that has incited real-world harm. The draft’s explicit attention to content amplification dynamics acknowledges this history without naming specific incidents — a careful but meaningful signal about legislative intent.

—

Points of Contention During the Comment Period

The 60-day comment window will surface several predictable fault lines. Platform companies are likely to push back on the 18-month audit cycle as operationally burdensome, particularly given how frequently recommendation systems are updated. Significant debate is also expected over who qualifies as an independent auditor, which methodologies are considered acceptable, and how commercially sensitive algorithmic information is protected during the audit process.

Civil society organizations and AI governance researchers, by contrast, may argue that the framework does not go far enough. The current draft does not mandate real-time transparency dashboards, pre-deployment algorithmic impact assessments, or user-facing explanations of why specific content was recommended — tools that researchers have identified as essential for meaningful accountability.

The ₹250 crore penalty ceiling will draw scrutiny as well. For large global platforms, that figure represents a fraction of annual revenue, raising legitimate questions about whether the financial deterrent is calibrated to change behavior or merely to signal regulatory intent.

—

What Comes Next

MeitY has invited responses from industry, civil society, academic institutions, and individual citizens through its official portal. The 60-day window closes the door on passive observation — every stakeholder with a position on AI accountability in India now has a formal opportunity to shape the final text.

For global platform compliance teams, the message is unambiguous: begin mapping your AI systems against this framework now, not after final rules are published. For AI governance researchers, the draft offers a concrete policy object to analyze, critique, and improve. For digital rights advocates, it represents the most substantive opening yet to embed meaningful transparency requirements into Indian platform law.

India’s draft AI accountability framework is imperfect, as all first drafts are. But it is serious, it is specific, and it is moving. The question is no longer whether India will regulate algorithmic systems — it is whether the final framework will be strong enough to match the scale of the problem it is trying to solve.