AI content labelling rules in India: a creator's guide

AI content labelling in India moved from a proposal to a rule this year. If you publish video, audio or images made with generative AI, two sets of rules now shape what you must do: India’s amended IT Rules, and the disclosure policy of the platform you upload to. This guide explains both in plain terms and ends with a checklist you can use on every upload.

It is written for creators and small studios, not lawyers. For anything high-stakes, read the primary sources linked below or ask a lawyer.

What changed in India in 2026

On 10 February 2026, the Ministry of Electronics and Information Technology notified an amendment to the IT (Intermediary Guidelines and Digital Media Ethics Code) Rules. It came into force on 20 February. It is the first Indian rule written specifically for what it calls synthetically generated information.

The definition covers audio, visual or audio-visual content that is created or altered by a computer “in a manner that appears real and authentic”. Routine editing is outside it: colour correction, noise reduction, compression, translation and accessibility improvements. This legal summary sets out the definitions and duties clearly.

The duties fall mainly on platforms:

• Labels and provenance. Permitted synthetic content must be prominently labelled and carry metadata or provenance markers that identify it.
• Declarations at upload. Large platforms, those with more than five million registered users in India, must ask users to declare whether content is synthetic, and use technical measures to check that declaration.
• Faster takedowns. Unlawful content must be removed within three hours of a government or court order, down from 36 hours.

So the rule rarely names creators directly. It reaches you through the upload form. When a platform asks whether your video contains realistic AI content, that question is how the law is being applied.

What YouTube asks creators to disclose

YouTube has had a creator disclosure since 2024. Its policy on altered or synthetic content asks you to tick a box in YouTube Studio when a viewer could easily mistake the content for a real person, place, scene or event.

Examples that need disclosure:

• A realistic face swap, or a real person shown saying or doing something they did not.
• A synthetic version of a real person’s voice.
• Altered footage of a real event or place.
• A realistic scene of a fictional major event, such as a disaster that never happened.

Examples that do not:

• Clearly unrealistic or animated content, and special effects.
• AI used for production help: scripts, ideas, captions, an AI voiceover on its own.
• Minor edits such as colour grading, beauty filters or background blur.

When you disclose, YouTube adds a label in the expanded description. For sensitive topics like health, news, elections and finance, the label appears on the player itself.

How the two fit together

Read side by side, the Indian rule and YouTube’s policy point the same way. Both care about realism and the chance of misleading someone. Neither is aimed at a creator who used AI to draft a script or to make an obviously animated short.

The practical rule of thumb: if a reasonable viewer could believe it really happened, disclose it. If it is plainly animation, fantasy or stylised art, a label is not required, though a plain line in the description is good practice.

There are two areas where we would be stricter than the minimum.

Real people

Never make realistic synthetic media of a real, identifiable person without their consent. This is where the legal risk sits, and it is also where audience trust breaks fastest. The three-hour takedown rule exists largely for this case.

News-like formats

Documentary and explainer formats borrow the look of journalism. If you recreate a real event with generated footage, say so clearly in the video, not only in the metadata. Viewers of news-style content expect footage to be real unless told otherwise.

A disclosure checklist for every upload

We build generative media pipelines as part of our engineering work, and the checks that last are the ones built into the process rather than remembered at upload time. Here is a version any creator can adopt:

1. Classify at the script stage. Mark each scene as real footage, stylised AI, or realistic AI. Decide disclosure before you generate, not after.
2. Keep a source log. Note which tools made which shots. If a platform or viewer asks, you can answer in a minute.
3. Answer the upload declaration honestly. If any scene is realistic AI, tick the box. The label is small. A strike or removal is not.
4. Add a plain description line. For example: “Some scenes in this video are AI-generated.” It helps viewers and costs nothing.
5. Do not strip provenance data. Some tools embed content credentials. Re-encoding can remove them, so keep the originals.
6. Check likeness and voice. No real person’s face or voice without written consent.

The takeaway

Labelling is becoming a normal part of publishing, the way thumbnails and captions are. Treat it as a production step, decided at the script stage, and it takes a few seconds per video. Treat it as an afterthought, and it becomes a risk to the channel.

If you run a studio and want disclosure and provenance built into your media pipeline, talk to us. We build review queues and checks like this into AI automation pipelines. For how we approach AI products more broadly, see our note on shipping software with AI agents.

FAQ

Do India's AI content labelling rules apply to individual creators?

The duties in the 2026 amendment fall on intermediaries, meaning the platforms. Platforms with more than five million users must ask uploaders to declare synthetic content and label it. In practice creators meet the rules through that declaration, so answer it honestly every time.

Do I need to label a video that uses an AI voiceover or an AI-written script?

On YouTube, not by itself. YouTube exempts AI voiceovers, scripts and captions used as production help. Disclosure is needed when realistic content could be mistaken for a real person, place or event.

Is animated or clearly fantastical AI video synthetic content under the rules?

Generally no. MeitY's approach and YouTube's policy both focus on content that looks real and could mislead. Obvious animation or fantasy is outside the main duty, though adding a plain note in the description costs nothing.

What happens if I do not disclose?

Platforms can apply labels themselves, limit or remove content, and act against accounts. Unlawful synthetic content, such as a deepfake of a real person, can also be taken down on a government or court order within three hours.