Deepfake Labelling Rules India: IT Rules 2026 on AI Content

Under the deepfake labelling rules India brought into force on 20 February 2026, platforms that let users create or share AI-generated audio, images or video must label such “synthetically generated information” clearly and prominently, embed provenance metadata where technically feasible, and never let labels be stripped. Takedown deadlines also shrank: 3 hours for a court order or lawful government intimation (earlier 36 hours), and 2 hours for a complaint about intimate or impersonating content, including morphed images (earlier 24 hours).

What counts as a deepfake under the 2026 rules

The amendment uses the term “synthetically generated information”: audio, visual or audio-visual information created, generated, modified or altered by a computer resource in a way that appears real or authentic and is likely to be seen as indistinguishable from a real person or real event.

It does not cover routine or good-faith editing, formatting, transcription, translation, accessibility improvements, educational or training material and research outputs, so long as they do not create false or misleading records. Colour-correcting a product photo is not a deepfake. A cloned voice of your managing director announcing a scheme is.

The amendment also makes clear that references to “information” used for unlawful acts include synthetic information, so every existing content rule now reaches AI output too.

What the deepfake labelling rules India notified require of platforms

• Labels: intermediaries that enable creation or sharing of synthetic content must ensure it is clearly and prominently labelled as synthetically generated. Reports of the final text describe on-screen labels for visuals and audio disclosures for sound.
• Provenance: where technically feasible, permanent embedded metadata or provenance markers, including a unique identifier of the computer resource used.
• No stripping: platforms must not enable removal, suppression or alteration of labels or metadata.
• Significant social media intermediaries: must ask users to declare whether content is synthetic before publication, use reasonable technical measures to verify the declaration, and show confirmed synthetic content with a clear notice.
• Safe harbour protected: removing synthetic content in compliance with the Rules does not breach the conditions of s.79 of the IT Act.
• User reminders: platforms must remind users of their rules and the consequences of breach at least once every three months.

An earlier amendment, notified on 22 October 2025 and in force from 15 November 2025, requires government takedown intimations to come from an officer not below Joint Secretary rank (or Director where no Joint Secretary is appointed) and police intimations from an officer not below Deputy Inspector General rank, with reasons and monthly review.

What it means for brands, creators and offices

1. Label your own synthetic content. Under the deepfake labelling rules India now enforces, large platforms will ask you to declare it. A false declaration can lead to removal and account action, and a realistic AI advert passed off as real may be a misleading advertisement under the Consumer Protection Act, 2019.
2. If your app generates avatars, voice-overs or edited photos, assess whether you are an intermediary enabling synthetic content, and build labels and metadata into the product.
3. Never use a real person’s face or voice without written consent. The Delhi High Court has protected personality rights against AI misuse, for example in Anil Kapoor v. Simply Life India (2023).
4. Keep a deepfake response plan ready, with named people, letters of authority and the platform reporting links.

A deepfake of your leader: what to do, in order

On a Saturday morning a video of the managing director of a Surat textile company appears online, “promising” guaranteed monthly returns to anyone who transfers money to a stated account. The communications head:

1. reports each URL to the platform as impersonation through synthetic content, with a letter of authority from the managing director, triggering the 2-hour clock;
2. files on cybercrime.gov.in, or calls 1930 if anyone has lost money;
3. posts a clear denial on the company’s official handles and alerts its bank;
4. escalates to the platform’s Grievance Officer, and then to the Grievance Appellate Committee at gac.gov.in within 30 days if the complaint is rejected.

Criminal law sits behind this: s.66D of the IT Act punishes cheating by personation using a computer resource, ss.67 and 67A cover obscene and sexually explicit material, and the Bharatiya Nyaya Sanhita, 2023 applies to the underlying fraud. A morphed image of a colleague shared in an office group can also be workplace sexual harassment for the Internal Committee under the POSH Act, 2013.

Frequently asked questions

What are the deepfake labelling rules India has notified?

They are amendments to the IT Rules, 2021, notified on 10 February 2026 and in force from 20 February 2026. Platforms that enable AI-generated audio, images or video must label it prominently, embed provenance metadata where feasible and prevent removal of labels, and large social media platforms must collect user declarations.

How fast must a platform remove a deepfake in India?

Within 2 hours of a complaint by or for the person shown, where the content is intimate or impersonates them, including morphed images. Content covered by a court order or lawful government intimation must go within 3 hours.

Do I have to label AI-generated images I post?

The legal duty to label falls mainly on platforms, but large platforms will ask you to declare synthetic content and a false declaration can get it removed. Labelling realistic AI images yourself is the safe practice for any business or creator.

Where do I complain if a platform ignores my deepfake complaint?

Escalate to the platform’s Grievance Officer and, if it rejects or ignores you, appeal to the Grievance Appellate Committee at gac.gov.in within 30 days. Where money or a crime is involved, also report on 1930 or cybercrime.gov.in.

Quick quiz

Build your response plan with AI Law and Governance in India, ₹599. Creators should also see Law for Content Creators and Influencers, ₹399.

Sources

• Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, as amended by notifications of 22 October 2025 and 10 February 2026; Rules 3(1)(d) and 3(2)(b)
• Information Technology Act, 2000, ss.66D, 67, 67A and 79
• Bharatiya Nyaya Sanhita, 2023; Consumer Protection Act, 2019; POSH Act, 2013
• Anil Kapoor v. Simply Life India, Delhi High Court, 2023

This page is for education and general information, not legal advice for a specific case. Law as on 31 October 2026.