On 8 October 2026, MeitY's Cyber Laws Division issued an advisory (File No. 16(2)/2026-CyberLaws) to all social media intermediaries. It asks them to apply "enhanced diligence" to false, misleading or manipulated content that could affect public order. It also asks them to "accord priority" to complaints from "authorised government agencies" and to "identify and appropriately label/contextualise manipulated, synthetically generated or misleading content," especially when it is presented as a current event. It took effect "with immediate effect" and landed about a week after protests in Delhi over the revision of electoral rolls, according to MediaNama.
The strongest case for the advisory
The government's concern is real. Old footage recirculated as breaking news is among the most common ways a tense public event turns into a law-and-order problem, and it needs no AI to work. An advisory that asks platforms to add context to recycled media and to answer police and ministry reports quickly is, on its face, modest. It creates no new offence, and it says action must respect "lawful expression and legitimate public discourse" and stay consistent with the IT Act, 2000 and the IT Rules, 2021, as reported by ThePrint. Context labels are also a lighter tool than removal, and many free-speech advocates prefer them.
Where the advisory goes further than the law
The problem is what the advisory leaves undefined, and what it quietly widens.
Labelling now covers more than the February rules. MeitY amended the IT Rules on 10 February 2026, in force from 20 February. The labelling provisions apply to "synthetically generated information" (SGI), which the Rules confine to audio, visual or audio-visual content that appears real, so text is outside them. MediaNama reports that the advisory's labelling clause carries no such limit and does not mention the amendment. The notified definition was already narrower than MeitY's October 2025 draft, which reached anything that "reasonably appears to be authentic or true." The advisory reopens that wider scope through guidance rather than through rulemaking. The Vikaspedia entry for the 2026 Amendment Rules is the public reference for that notified text.
"Misleading" is a judgment call handed to private companies. Synthetic media can be tested against provenance signals. "Misleading" cannot. A photo can be old and still illustrate a real grievance, and a satirical clip can be accurate in substance. Platforms that are told to act "with immediate effect" and are not told what the standard is will resolve doubt by labelling or removing more, not less.
"Authorised government agencies" is undefined. ThePrint notes that the advisory does not say who counts. Nor does it say what "priority" means in practice, or how "enhanced diligence" differs from the due diligence the Rules already require. MediaNama adds that significant social media intermediaries already carry additional due-diligence duties under Rule 4. A priority queue for unnamed agencies, attached to a public-order rationale, invites the pattern in which complaints from any official office are treated as binding.
It sits uneasily with Shreya Singhal. In Shreya Singhal v. Union of India (24 March 2015), the Supreme Court read down the "actual knowledge" trigger in Section 79(3)(b) so that an intermediary must act on a court order or a government notification, and that either "must strictly conform to the subject matters" of Article 19(2) (para 117). The judgment's reading of the then-current Rule 3(4) went further and tied knowledge to a court order alone (para 118). The full text is on Indian Kanoon. "Potential implications for public order" is an Article 19(2) ground, but an advisory that tells platforms to give weight to agency complaints, without saying whether those complaints are formal Section 69A or court-backed directions, blurs the line the Court drew. Platforms may reasonably read an informal request as something they ignore at the risk of safe harbour.
There are no stated penalties, but that is not reassuring. ThePrint reports that, unlike MeitY's advisories after the November 2025 Red Fort blast and on obscene content in December 2025, this one specifies no consequences. That makes the advisory harder to challenge, because nothing is formally imposed, while leaving the risk implicit. Companies know that earlier advisories tied non-compliance to the loss of Section 79 protection, and they will price that in.
A narrower, better-designed path
The goal of reducing recycled-media confusion during unrest is achievable without these costs:
- Put it in the Rules or drop it. If labelling of non-synthetic misleading content is to be required, it should go through notified rulemaking with consultation, as the SGI provisions did, not through an advisory.
- Define the requesting bodies. Publish the list of authorised agencies, the legal provision each acts under, and a standard format for requests.
- Prefer context to removal and keep a record. Labelling with a link to a fact-check or the original date is proportionate. Platforms should be asked to publish aggregate counts of agency-prioritised actions so the public can see how the channel is used.
- Protect text and commentary. Opinion, satire and reporting about the protests should be expressly outside any "misleading" standard.
India's own rulemaking record shows the government can draw a defensible line: it narrowed the October 2025 SGI draft after industry objections, including from Nasscom, which argued that much internet content would need labels. The advisory undoes some of that discipline in a single page, at the moment when the stakes for political speech are highest.
The pro-innovation position is not that platforms should ignore manipulated content. It is that duties which shape what 1.4 billion people can see should be written down, bounded and reviewable. A week-old protest is a poor reason to skip that process.