Platform Power, Misinformation and the Unfinished Business of Speech Regulation

Every election cycle in India now produces the same ritual complaint: platforms are flooded with doctored videos and fabricated quotes, the Election Commission issues advisories, and social media companies respond with a mix of automated takedowns and public relations statements. What rarely gets examined with the same intensity is the legal architecture underneath this ritual, an architecture built in stages since 2000 and now straining under demands nobody anticipated when Section 79 of the Information Technology Act first offered platforms a conditional safe harbour from liability for user content.

The safe harbour bargain

Section 79 was, and remains, a reasonably sensible compromise. It shields intermediaries such as social media companies, app stores and internet service providers from liability for content posted by users, provided they act as neutral conduits and remove unlawful content upon receiving actual knowledge of it, typically through a court order or a government notification. This is the same basic bargain that underpins Section 230 of America's Communications Decency Act, and for good reason: without it, no platform could function at the scale Indian users now expect, since manually vetting billions of posts before publication is simply not possible. The bargain's central tension, though, is that "actual knowledge" and the process for acting on it have been defined and redefined by rules, not by Parliament, and each redefinition has shifted power toward the executive.

The 2021 rules and their aftermath

The Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, changed the texture of platform regulation considerably. They introduced traceability requirements for messaging services, mandatory grievance officers based in India, a three-tier regulatory structure for digital news and OTT content, and tighter takedown timelines. WhatsApp challenged the traceability mandate in the Delhi High Court, arguing it would break end-to-end encryption for over 400 million Indian users, a case that remains unresolved years later and illustrates how quickly technical architecture and legal drafting can collide. The rules also created, for online news, a self-regulatory structure overseen ultimately by an Inter-Departmental Committee under the Ministry of Information and Broadcasting, a body with no counterpart in how print media has historically been regulated in India, and that asymmetry between old and new media rules is itself worth scrutiny.

The fact-check unit episode

The most instructive recent episode is the 2023 amendment that empowered a government-notified Fact Check Unit to flag as "fake, false or misleading" any online content relating to the business of the central government, with platforms expected to remove such content or lose safe harbour protection. The Bombay High Court, in a split verdict eventually resolved by a third judge, struck this provision down in 2024, holding that a unit within the government could not be judge in its own cause when the content being assessed concerned the government's own conduct. This was an important corrective, not because misinformation about government policy is a trivial problem, but because the remedy proposed inverted a basic principle of natural justice. The judgment did not deny that fact-checking has a legitimate place in combating misinformation; it insisted only that the checker cannot also be the accused party's advocate.

What takedown data actually shows

Transparency reports published voluntarily by major platforms, along with data periodically disclosed in Parliament, suggest that India issues among the highest volumes of content-blocking directions of any democracy, running into tens of thousands of URLs annually under Section 69A of the IT Act. What these reports rarely reveal is the reasoning behind individual blocking orders, because Section 69A proceedings are explicitly confidential, a design choice defended on security grounds but one that leaves affected users and even the platforms themselves unable to contest specific orders on their merits. The Karnataka High Court's ruling in the X Corp case, upholding a broadened blocking mechanism called Sahyog, has reopened this debate, with digital rights groups arguing that a system without published reasons or a meaningful appellate stage cannot satisfy the proportionality standard the Supreme Court set out in Shreya Singhal v Union of India, the 2015 judgment that struck down the notorious Section 66A.

The asymmetry that needs fixing

The honest tension in this space is that removing genuinely dangerous content, from deepfaked communal provocations to coordinated election disinformation, sometimes needs to happen fast, and courts move slowly. Nobody serious argues that every takedown decision should await prolonged litigation while a fabricated video showing a political leader making incendiary remarks circulates unchecked. But speed and accountability are not mutually exclusive; they are simply expensive to build together. A takedown order could be issued quickly and still be published in redacted form afterward, with a genuine appellate mechanism, ideally judicial rather than purely executive, available to the platform and to the affected user. Several democracies, including through the EU's Digital Services Act, have tried to build exactly this kind of layered accountability, requiring large platforms to publish detailed transparency reports and offer users a right to explanation for content removal decisions.

India's regulatory instinct, by contrast, has consistently favoured discretion over disclosure, understandable in a country that genuinely faces communal fault lines and disinformation actors, both domestic and foreign, operating at scale. But discretion without disclosure erodes public trust in exactly the mechanism meant to protect it. When citizens cannot tell whether a blocked post was removed because it incited violence or because it embarrassed an official, they reasonably suspect the latter, and that suspicion, once it takes root, is far harder to fact-check than any single piece of misinformation.