Two platform duties in ten days
Between 30 September and 8 October 2026, the obligations sitting on Indian online platforms moved twice. Neither change is a gazetted rule yet in the sense of a finished, enforceable text, and that distinction matters for how a startup should read its insurance programme. What has changed is the direction of travel, and the direction is clear enough that a renewal in the next two quarters should be written against it.
The first move came in the Supreme Court. As Inc42 reported on 30 September, the Solicitor General told a bench of CJI Surya Kant, Justices Joymalya Bagchi and V Mohana that the government would amend the IT Rules to make it a statutory obligation for intermediaries not to let minors hold social media accounts. The second came from the executive. MeitY's Cyber Laws Division issued an advisory dated 8 October 2026 asking social media intermediaries to label or contextualise "manipulated, synthetically generated or misleading content", with immediate effect (MediaNama, 8 October 2026).
The scale is large. Inc42 Datalabs estimates over 655 million social media users in India. Most of the attention will go to the global platforms, but the duties attach to intermediaries generally, and a long tail of Indian consumer-internet, creator, community, dating and gaming startups host user content and user accounts. For those companies the questions are practical: who sends the notice, who pays to answer it, and which policy, if any, picks up the cost.
This post builds on our earlier analysis of the notified synthetic-media rules and which insurance tower answers an unlabelled AI asset. That piece mapped media liability, technology E&O, cyber and D&O against a labelling failure. This one looks at what the two new duties add, and at how a startup's smaller, less negotiated programme responds.
The under-18 bar: a stated intention, not an amended rule
Read the statement to the court for what it is. The government told the court it would amend the IT Rules. As of this writing, no amended text has been notified, no age threshold mechanics have been published, and no verification standard has been specified. A founder who tells the board that India has banned under-18s from social media is overstating the position, and a broker who prices it as an in-force obligation is doing the same.
That said, an intention stated to the Supreme Court by the government's second law officer is not a casual signal. It tells you three things a risk manager can act on now:
- The duty will sit on the intermediary. The framing was a statutory obligation on intermediaries not to allow minors to hold accounts. That places compliance, and the consequence of failing it, on the platform rather than on parents or users.
- It will travel through the IT Rules. The same instrument governs due diligence and safe harbour under Section 79 of the IT Act. A failure to comply is therefore likely to be framed as a loss of safe harbour plus regulatory action, not only as a standalone penalty.
- Account-level controls will be scrutinised. Whatever the final text says about verification, a platform that cannot show what it did to keep minors off will be in a weak position in any inquiry that follows.
The category of startup most exposed is the one that never thought of itself as social media: a gaming app with chat, an edtech product with peer forums, a fan community, a short-video creator tool, a dating or friendship app with lax onboarding. If users create accounts and post content others can see, the company should assume it will be asked how it keeps minors out.
The misleading-content advisory: wider than synthetic media, and undefined
The 8 October advisory is the more immediate change because it took effect on issue. It extends the labelling logic that the synthetic-media rules introduced to two further categories: content that is manipulated and content that is misleading. Synthetic content was at least a technical category; a model produced it or it did not. Misleading content is a judgement.
MediaNama's analysis on 8 October identified two gaps. The advisory does not define the "authorised government agencies" whose flags platforms are expected to act on, and it does not set standards for what counts as misleading. The effect, in MediaNama's reading, is to leave platforms to make that call themselves.
For an insurance buyer, that gap is the whole point. Undefined standards produce two opposite claim shapes, and a startup can face both in the same quarter:
- Under-labelling. A user posts something that a regulator, a complainant or a court later treats as misleading. The platform did not label or contextualise it. The exposure is regulatory (a MeitY notice, a question over safe harbour) and civil (the person harmed by the content sues the platform as well as the poster).
- Over-labelling. The platform labels a creator's or a business's content as misleading, or contextualises it in a way the creator disputes. The creator alleges defamation, loss of income or breach of the platform's own terms. The exposure is a media-content claim and a contract claim, brought by someone the startup depends on.
A startup cannot avoid both by policy design. What it can do is document the rule it applied, apply it consistently, and make sure its insurance responds to the claim shape it is most likely to see.
The Meta inquiry shows how the notices arrive
The clearest preview of the process was reported on the same day as the advisory. After a BBC investigation into Instagram ads leading to CSAM channels, MeitY summoned Meta and the NCPCR opened an inquiry (Inc42, 8 October 2026). Two agencies, two separate proceedings, triggered by a media investigation rather than a complaint.
Scale that down to a startup and the pattern holds. The likely sequence for a mid-sized Indian platform is:
- A journalist, an activist group or a parent surfaces content involving minors, or content that is plainly misleading, on the platform.
- MeitY issues a notice or summons asking what the platform did and why.
- A child-rights body such as the NCPCR opens its own inquiry, with its own document demands and hearings.
- Users, parents or affected businesses file civil suits or consumer complaints, sometimes coordinated.
- Investors ask the board what it knew, and when.
Each step has a cost, and the costs land on different policies, or on none. The startup that has priced its insurance only against a data breach is covered for none of the first four steps, and possibly not the fifth.
Investigation and inquiry costs are often the largest part of a regulatory event for a small company: outside counsel, a forensic review of moderation logs, executive time at hearings. Many startup programmes cover these only when a formal "claim" exists, and a summons or an advisory-driven query may not meet the policy's definition.
Which cover responds: MeitY notice, NCPCR inquiry, user class action
Take the three events in turn and the policies that might answer each.
A MeitY notice or an NCPCR inquiry
The responding tower, if any, is D&O for the individuals and, sometimes, cyber for the entity's regulatory defence. Most Indian startup D&O wordings define a claim to include a formal investigation or proceeding against an insured person by a regulator, and many add pre-claim inquiry cover for costs incurred when an officer is asked to attend or produce documents. The gap is entity cover: outside securities claims, a notice addressed to the company is often outside D&O. Cyber regulatory cover usually responds only where the proceeding arises from a security or privacy failure, so a notice about labelling or minors' accounts typically falls outside it unless a data-protection angle is present. Check whether the definition of regulator reaches MeitY acting under the IT Act, rather than only financial regulators.
An inquiry by a child-rights commission such as the NCPCR is, for coverage purposes, another regulatory investigation. The same D&O questions apply. Two additional points arise. First, any allegation touching sexual content involving children will meet conduct and illegality exclusions very quickly, and insurers will be cautious about advancing defence costs. Second, if the inquiry concerns harm to a specific child, the facts may also support a bodily injury framing (psychological harm is commonly pleaded), which many professional and media wordings exclude.
A user class action or coordinated complaints
Here the claim is from third parties and the natural responders are media liability (for content claims such as defamation, misleading content or infringement), technology E&O or professional indemnity (for a failure of the service the platform promised), and cyber (if the complaint is really about personal data, including children's data processed without valid consent). Our media liability insurance guide covers the base product; the question for a platform is whether it covers user-generated content it hosts as well as content it publishes itself. D&O enters only if the class action, or a follow-on investor claim, names directors.
The exclusions that usually bite
Startup programmes are often bought quickly, on standard wordings, with little negotiation. Three exclusions do most of the damage in this kind of event.
Regulatory fines and penalties. Almost every Indian liability wording covers defence costs but excludes fines and penalties, or covers them only "where insurable by law". Even where a wording is silent, insurability of a regulatory penalty in India is uncertain, and an insurer will usually contest it. Buyers should assume that defence is the realistic insurable component, and size limits to cover sustained defence through a multi-agency inquiry rather than the eventual penalty.
Bodily injury. Media liability, technology E&O and cyber typically exclude bodily injury, sometimes with a carve-back for mental anguish arising from a covered media wrong such as defamation. Claims involving minors are frequently framed around psychological or physical harm. If the carve-back is narrow, a claim pleaded as harm to a child may fall between the liability wording (which excludes bodily injury) and public liability (which was written for physical premises, not online platforms). Read the definition of bodily injury, and ask whether mental anguish or emotional distress is carved back across all wrongful acts or only for defamation.
Intentional, dishonest or deliberate acts. Every tower excludes deliberate wrongdoing, but the trigger varies. A good wording applies the exclusion only after a final adjudication or written admission, and severs it so that one officer's conduct is not imputed to others. A poor wording lets the insurer decline on allegation. For the misleading-content advisory, where a platform will be making deliberate labelling decisions every day, the exclusion should turn on knowledge of wrongdoing, not on the fact that the platform intentionally chose to label or not label.
Where the exposure concentrates by startup type
The two duties do not hit every platform the same way. A short guide to where each category should look first:
- Consumer social and community apps. The under-18 bar, if notified, applies to them directly. Their main exposure is regulatory and D&O, plus content claims from both directions under the advisory. Priority: D&O with pre-claim inquiry cover, media liability that names user-generated content.
- Creator and influencer platforms. The advisory's over-labelling risk is sharpest here, because creators earn from the reach a label can reduce. Priority: media liability and E&O that respond to claims by creators, not only by the public. Our creator economy insurance guide covers the base programme.
- Gaming and esports platforms. In-game chat, team communities and a young user base put them squarely in scope of both duties. See our gaming and esports platform insurance post for the wider programme.
- Edtech with peer or forum features. The user base is often largely minors by design, which makes an account bar on social features an operational question rather than a theoretical one. The edtech platform insurance guide covers the data and professional exposures that sit alongside.
In each case, the misleading-content advisory adds a moderation-decision exposure that most startup E&O wordings were not written for. The wording usually covers failure to perform the service; it may not contemplate a claim that the platform performed a judgement call badly.
What to do before the amendment lands
Use the window between a stated intention and a notified rule to fix the programme rather than wait for the text.
- Map the claim paths in writing. For a MeitY notice, an NCPCR inquiry and a user class action, list which policy responds, under which insuring clause, and whether a formal "claim" is required. Send it to your broker and ask for confirmation by endorsement.
- Fix the D&O definitions. Ask for pre-claim inquiry cover, a definition of regulator that includes MeitY and statutory child-rights bodies, and, where available at your stage, an entity investigation extension.
- Confirm user-generated content in media liability. Check that hosted content, not only published content, is covered, and that claims by your own creators or users are not excluded as contractual.
- Read the minors and sexual-content exclusions. If one exists, negotiate it to apply only to knowing conduct, or price the gap honestly to the board.
- Document the labelling rule. Write down how the platform decides what is misleading or manipulated, who decides, and how disputes are handled. Under an advisory that leaves the standard undefined, a consistent written rule is your best defence, and it is what an insurer will ask for at the next renewal proposal.
- Prepare an age-assurance record. Whatever the final rule requires, be able to show what you did to keep minors off account-holding features, and when.
None of this requires the amendment to be notified. All of it makes the eventual notice cheaper to answer.