FILE IGRS-122 · AI & LAW · 16 NOVEMBER 2026

India's Deepfake and AI-Content Rules (IT Amendment Rules 2026): Labels, Metadata, 3-Hour Takedowns and What It Means for Investigators

By Manuj Kumar Pandey16 November 2026 AI & Law AI Labels 3-Hour Takedown In Force 20 Feb 2026
India's Deepfake & AI-Content Rules 2026 — IGRS

India now has specific rules for AI-generated content. The Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, 2026 were notified by the Ministry of Electronics and Information Technology on 10 February 2026 and came into force on 20 February 2026. They introduce the legal category of “synthetically generated information” and put new duties on platforms. This file explains what changed, what platforms must do, and how victims and investigators can use the rules.

What Counts as “Synthetically Generated Information”

The rules cover audio, visual and audio-visual content that is “artificially or algorithmically created, generated, modified or altered using a computer resource” so that it appears real or authentic — in particular content likely to be perceived as indistinguishable from a real person or a real event. Good-faith routine editing and formatting, document creation, accessibility features such as translation, and improvements to clarity are carved out, as analysed by King Stubb & Kasiva.

What Platforms Must Now Do

ObligationWhat it means
Prominent labelsAI-generated content must be labelled so users can immediately identify it as synthetic (with audio disclosure where relevant)
Permanent metadataEmbedded metadata or provenance markers, including a unique identifier, to trace the generating or modifying computer resource
No strippingPlatforms must not allow labels or metadata to be removed or suppressed
User declaration (social media)Users must declare whether content is AI-generated before publishing; significant social media intermediaries must deploy technical measures to verify declarations
Prohibited synthetic contentAI-generated child sexual abuse material, non-consensual intimate imagery, false documents and misleading depictions of real people or events
Consequences for usersRemoval, suspension or termination of accounts, disclosure of the violator's identity to victims, and reporting to law enforcement

Faster Timelines

ActionBeforeNow
Acting on certain lawful orders36 hours3 hours
Responding to user grievances15 days7 days
Certain grievance actions24 hours2 hours

Timelines as reported by Business Today and King Stubb & Kasiva. Platforms that knowingly allow violating content risk being treated as failing due diligence, which puts their safe-harbour protection under Section 79 of the IT Act at risk.

Using the Rules: A Victim's and Investigator's Workflow

#Step
1Preserve first: capture the post, profile and URL with date and time visible; download the media; hash each file (SHA-256)
2Report to the platform's grievance officer, citing the IT Rules and the specific category (for example non-consensual intimate imagery or impersonation)
3File at cybercrime.gov.in or call 1930; for intimate imagery, request urgent removal
4Examine the media for labels and embedded metadata; their absence on a platform that should apply them is itself relevant
5For investigators: request uploader account, IP and device records through a notice under S.94 BNSS 2023
6Produce electronic records with a S.63 BSA 2023 certificate

Analysis: The metadata requirement matters most for investigators. Provenance markers and unique identifiers create a new trail from a piece of content back to the tool or service that generated it. Expect bad actors to use tools outside these rules or to re-encode content to strip markers — so the absence of metadata is not proof that content is real.

Applicable Offences

ProvisionTypical deepfake scenario
S.66C IT Act 2000Using another person's identity features, such as face or voice data, fraudulently
S.66D IT Act 2000Cheating by personation using a computer resource — deepfake video calls and impersonation scams
S.66E IT Act 2000Publishing images of a person's private area without consent
S.67 / S.67A IT Act 2000Publishing obscene or sexually explicit material electronically
S.318 / S.319 BNS 2023Cheating; cheating by personation
S.79 IT Act 2000Platform safe harbour, conditional on due diligence under the IT Rules

How Dunime Can Help

This section describes the publisher's own company. Disclosure: IGRS is written and maintained by Manuj Kumar Pandey, founder of Dunime Technologies.

NeedDunime offering
An organisation impersonated by a deepfake of an executive or brandIncident response support and post-incident forensic audits
Reviewing how a business collects and handles images, voice and video of real peopleCyber audits including DPDP Act data-protection reviews

Frequently Asked Questions

What are India's new deepfake rules?
The Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, 2026, notified by MeitY on 10 February 2026 and in force from 20 February 2026. They regulate 'synthetically generated information' — AI-created or altered audio, images and video that appear real — through labelling, metadata, user declarations and faster takedowns.
Do AI-generated images and videos have to be labelled in India?
Yes. Intermediaries that enable AI-generated content must label it prominently so users can immediately identify it as synthetic, embed permanent metadata or a unique identifier, and must not allow labels or metadata to be removed.
How fast must platforms remove unlawful deepfakes?
As reported by Business Today, the time to act on certain lawful orders was cut from 36 hours to 3 hours, user grievances must be answered within 7 days instead of 15, and certain grievance actions must be completed within 2 hours.
How do I report a deepfake of myself in India?
Report it to the platform through its grievance mechanism, file a complaint at cybercrime.gov.in or call 1930, and preserve the content with its URL, date and time before it is removed.

Sources

Facts in this file are drawn from the public reporting below. Where this file goes beyond the sources, the text is marked as analysis or inference.

Related Resources