India's Ministry of Electronics and Information Technology (MeitY) has told Meta to make sure its recommendation systems on Facebook and Instagram do not expose users to "a single or limited set of narratives." According to MediaNama's report, which draws on PTI, the advisory is dated 10 October 2026 and was sent to Meta alone. It asks Meta to act "with immediate effect", to "endeavour to provide appropriate diversity of relevant and credible information", and to report back on what it does. MeitY has not published it.
The strongest case for the ministry
The concern behind the advisory is not frivolous. Recommendation systems are not neutral pipes. They decide what millions of people see during fast-moving political events, and engagement-optimised ranking can reward outrage and amplify misleading images. Ministry sources told PTI that agencies had reported Meta's algorithms appeared to favour one kind of content over others. A government that asks a platform to add context to images of ongoing events, or to slow the spread of misleading material, is asking for something many researchers and platform trust-and-safety teams would call reasonable in the abstract. Governments are entitled to raise these concerns with companies.
Why this instrument fails
The problem is the form, the vagueness and the target, not the aspiration.
It has no statutory footing. MediaNama reports that the advisory cites the IT Act and the due diligence obligations under the IT Rules. It also reports that neither sets rules for recommendation systems. Internet Freedom Foundation (IFF) founder Apar Gupta told PTI the advisory is not legally binding and is not tied to any provision of the IT Act or IT Rules. A document that binds no one, rests on no provision and is shown to no one can still change what a platform does, because the platform fears the consequences of ignoring it.
It is unpublished. A rule that cannot be read cannot be challenged, and a company cannot tell whether it is complying. The Supreme Court of India has been clear that restrictions on speech must be reasoned and reviewable. In Anuradha Bhasin v. Union of India (10 January 2020), the Court applied a proportionality test with four stages: a legitimate goal, a rational connection to it, necessity (no less restrictive alternative), and a balancing of impacts. It held that authorities must identify the goal and consider alternatives, and that restriction orders must rest on sufficient material and be open to judicial review. It also held that free speech over the internet is protected under Article 19(1)(a). A confidential, open-ended advisory cannot be tested against any of these steps.
Its key terms cannot be applied. PTI's report, as relayed by MediaNama, does not say how Meta should judge which narratives are "credible", or how much exposure to one narrative counts as disproportionate. Gupta pointed to similarly vague phrases, such as content "without adequate context" and periods of "heightened public-order sensitivity". A recommender that is told to avoid "limited" narratives while a single event dominates the news faces an impossible instruction. Does a protest that millions are watching get diluted with unrelated material? Does the platform demote the protest, or promote counter-narratives to balance it?
The political context makes the vagueness dangerous
The advisory was issued on the day of the Cockroach Janta Party (CJP) protests in Delhi, which demanded the resignation of Chief Election Commissioner Gyanesh Kumar. It does not name any incident. MediaNama notes that the CJP's Instagram account had 25.1 million followers as of 19 October, against the BJP's 10 million, and that the CJP account overtook the BJP's within four days of its launch in May.
These facts do not prove the advisory targets the CJP. The text, as reported, does not say so. But an instruction to diversify what users see, issued on the day of a viral protest against an election official, and enforceable only through the ministry's goodwill, invites exactly that reading. Whether or not it was intended, the effect is to make a platform's ranking choices depend on how the government reads a political moment.
The advisory also does not stand alone. IFF's site carries a statement dated 9 October 2026 criticising a separate MeitY "enhanced diligence" advisory. As IFF describes it, that advisory asks platforms to proactively monitor content around protests in Delhi and Mumbai. IFF argues it would pressure platforms to police protest speech and should be withdrawn. The two documents are distinct, and the Meta advisory is not named in IFF's summary. Together, however, they point to a pattern of non-statutory instructions during political events.
A proportionate alternative
If the state believes recommendation systems create public-order or misinformation harms, the proportionate route is public and legal. The Court in Shreya Singhal v. Union of India (24 March 2015), which concerned Section 66A of the IT Act, scrutinised vague internet-speech provisions against Article 19(2) standards. The same discipline applies here. A better approach would have four parts:
- Publish the advisory, or replace it with a consultation draft that sets out specific, testable obligations.
- Prefer transparency over outcome targets. Require platforms to disclose how their recommenders rank civic content, and let independent researchers audit the results, rather than prescribe which narratives should surface.
- Add context labels to images of ongoing events through clearly defined, neutral and appealable processes, not confidential instructions.
- Place any binding duty in statute or rules, so that it can be debated in Parliament and challenged in court.
Each of these addresses the genuine concern about amplification without handing the executive a discretionary lever over what an open platform shows its users.
The wider cost
Pro-innovation regulation depends on predictability. A company serving hundreds of millions of Indian users cannot build compliance systems around instructions it receives privately and that change with events. The incentive is to over-comply: to quietly reduce the reach of anything that might draw attention. That chills the speech the government says it wants to see more of. IFF's call for MeitY to withdraw the advisory, and for Meta to disclose every takedown and downranking it makes under it, is the minimum standard of accountability. Meta should publish the advisory, or at least its own response to it, and MeitY should explain on what legal basis it expects compliance.