Blog · Research ·
Do AI ads need a label in India?
Short answer
Some do. ASCI's new guidelines sort AI-made ad content into three groups: things that are banned even with a label, things that need a label such as “Video created using AI”, and things that need nothing. An AI presenter needs a label. A colour-corrected shot does not. A fake testimonial is not allowed at all.
If you have seen the news about AI labels and wondered whether your next Reel is about to land you in trouble, this is for you. It is based on ASCI’s published guidelines and on reporting about the IT Rules. It is a plain summary, not legal advice. If your ad makes medical, financial or similar high-stakes claims, talk to a lawyer.
Who made these rules?
Two separate things happened in 2026.
- ASCI, the Advertising Standards Council of India, published guidelines in September 2026 on labelling AI-made content in ads. They come into effect three months after publication, which puts them around the end of 2026. ASCI is a self-regulatory body, but its code is widely followed across Indian advertising.
- The government amended the IT Rules in February 2026. Those rules mainly put duties on platforms: they must label realistic AI-made content, and large social media platforms must ask users to declare whether what they upload is AI-made.
In practice, that means two questions for an advertiser. Does my ad need a label under ASCI? And when I upload it, am I answering the platform’s AI question honestly?
What counts as “AI-made” here?
Not everything AI touched. ASCI is talking about audio, images or video that are generated or materially changed by a computer to look real, and that show a person or an event in a way people could believe is real. A cartoon dragon is not the target. A realistic woman holding your serum is.
Group 1: banned, even with a label
A label does not rescue these. ASCI’s examples:
- Fake endorsements or testimonials, where AI makes it look like a real person used or recommended your product when they did not
- Exaggerated results, like an AI “after” photo of a weight-loss product that does not show what the product can really do
- Places that do not exist shown as if they do, such as a hotel or facility that is not real
- Deepfakes, copyrighted work used without permission, or anyone’s likeness without their consent
Group 2: allowed, with a label
These need a clear disclosure. ASCI’s examples include:
- AI influencers, presenters and brand ambassadors
- A real person’s face or voice recreated for personalised messages, even with their consent
- Made-up events or situations that could change how people understand what the product does
- A product that does not exist yet, like a 3D walk-through of an unbuilt apartment block
- Sound effects that are central to the product, like audio quality in an ad for a headset
- Paid product suggestions made by an AI, which must say “Sponsored by [Brand]”
Group 3: no label needed
ASCI is clear that it does not want labels on everything. No label is needed for:
- Routine edits: colour correction, light adjustment, noise reduction, standard blemish removal, formatting, copy polish
- Decorative backgrounds, skylines, ambient music, jingles and crowd sounds that say nothing about the product
- Obvious fantasy that nobody would take as real
- Using AI to write or polish ad copy
- Subtitles, captions, translations and audio descriptions for accessibility
What should the label say?
ASCI does not insist on one wording. It gives “Audio/Video created using AI” and “Audio/Video enhanced using AI” as examples, and also accepts the labels that ad platforms provide. Any label that tells the viewer the truth works.
Three questions to ask about your own ad
ASCI’s guidelines turn on three questions:
- Is the AI part central to how the ad persuades, or does it shape the buying decision?
- Without a label, would the AI part mislead people about what the product claims or does?
- Would leaving the label off give a false impression of the product’s benefits or performance?
If the answer to any of these is yes, label it.
What this means for an ad from us
Our creator-style ads use generated presenters, so they sit in Group 2 and need a label. A product ad that rebuilds your real pack with better lighting is closer to Group 3, unless it shows results the product cannot deliver. The things in Group 1 were never on the menu: we do not make fake testimonials, and we never present a generated person as a real customer.
One more thing worth saying plainly. A label does not make a false claim true. ASCI says any ad with AI content must still follow its whole code. You are responsible for the claims in your ad, and when a claim needs proof, we say so when we read your brief.
Want the bigger picture on how people react to AI ads? Read Can people tell your ad was made with AI?
Frequently asked questions
Do AI ads need a label in India?
Some do. Under ASCI's 2026 guidelines, realistic AI content such as AI presenters and influencers needs a label like “Video created using AI”, routine edits need none, and fake testimonials are not allowed even with a label.
When do ASCI's AI labelling guidelines take effect?
ASCI published them in September 2026, and they come into effect three months after publication, around the end of 2026.
Does colour correction or AI-written ad copy need a label?
No. ASCI does not require labels for routine edits such as colour correction and noise reduction, for decorative backgrounds and music, or for using AI to write or polish ad copy.