India's technology ministry has told Meta to stop its recommendation systems from funnelling users toward a narrow set of narratives, and to tighten safeguards against content that could spark unrest or incite violence. The advisory, dated October 10, lands against the backdrop of protests in Delhi over the Election Commission's handling of the Special Intensive Revision of electoral rolls, though the ministry has not linked the order to any single event.
What the Ministry Wants Changed
At the centre of the advisory is a concern that Meta's ranking algorithms are not neutral arbiters of what users see. Officials cited reports suggesting the company's systems were favouring particular kinds of content over others, raising the risk that users encounter a skewed picture of unfolding public events rather than a fair cross-section of credible information. The ministry wants Meta to audit its recommendation and content-ranking mechanisms so that misleading, manipulated or out-of-context material is not repeatedly pushed to the top of feeds.
A second demand concerns context. Photographs, videos and graphics tied to live public events should carry clear markers of authenticity, date and location, particularly when those details are not self-evident. The ministry singled out old or unrelated visuals repurposed as breaking news, along with unverified claims about police action, emergency measures or official restrictions, as recurring problems that platforms need to catch before they spread.
How Recommendation Engines Shape What People See
Social media platforms do not show users a neutral stream of posts in chronological order; they rank content using engagement signals, past behaviour and predictive models designed to maximise time spent on the app. This architecture rewards material that provokes strong reactions, which can mean emotionally charged or divisive content rises faster than measured, verified reporting. During periods of civil unrest, that dynamic can amplify rumours or selectively edited footage well before fact-checkers or moderators intervene, hardening public perception around a single, possibly distorted, version of events.
This is the mechanism MeitY is targeting: not individual posts in isolation, but the systemic tendency of ranking algorithms to concentrate attention on a limited slice of the information landscape. The advisory asks Meta to build in "appropriate diversity" so that recommendation systems do not become instruments of narrative capture during sensitive periods.
Legal Basis and Industry Pushback
The ministry grounds its request in the Information Technology Act, 2000, and the IT Rules, 2021, which require intermediaries to exercise due diligence and to inform users about prohibited categories of content, including material likely to incite a cognisable offence against public order. Meta has also been told to cooperate promptly with law enforcement requests within the timelines the rules prescribe, and to report back on steps taken.
Digital rights advocates are unconvinced. Apar Gupta of the Internet Freedom Foundation has described the advisory as a continuing censorship directive that lacks binding legal force, arguing that terms like "adequate context" and "heightened public-order sensitivity" are not anchored to any specific statutory provision. He has called on the ministry to withdraw it and on Meta to publish details of any takedowns or downranking carried out in response.
What Comes Next
Advisories of this kind sit in a grey zone: they are not court orders, yet platforms rarely ignore them given the regulatory leverage the ministry holds under the IT Rules. Meta now faces the practical challenge of demonstrating compliance without a precise legal yardstick for what counts as sufficient "diversity" in its recommendations. The episode adds to a broader pattern of governments worldwide pressing platforms to take greater responsibility for algorithmic amplification during periods of political tension, a trend likely to intensify as elections and protests increasingly play out as much on social feeds as in the street.
