ASCI's AI ad labels start in December: what creator deals must show

Indian brands now use AI for almost every creator campaign in some form: a cloned voice for a regional cut, an AI avatar for a quick explainer, a generated background behind a real influencer. From December, India's ad self-regulator expects many of those uses to be labelled on screen.

When do ASCI's AI labelling rules apply?

The Advertising Standards Council of India issued the final guidelines in September 2026 after consulting on a draft released in May. afaqs reported the release on 29 September and says the rules take effect three months after publication. A LexOrbis analysis gives the guidelines' date as 17 September 2026, and a DSCI brief also says September. Because the sources differ on the exact publication day, the start date falls somewhere between mid and late December 2026. We could not find ASCI's own PDF to settle it, so plan for the earlier date.

What has to be labelled?

The test is whether synthetic content could materially influence a consumer's decision and whether leaving out a label could mislead. Law firm Saraf and Partners and LexOrbis list the main cases:

• AI influencers or brand ambassadors, meaning virtual faces that promote a product.
• A real person's face or voice recreated with AI, even with their consent, for example a personalised message in a creator's cloned voice.
• Fabricated events or settings that change how people understand the product.
• Demos of products that do not exist yet.
• AI sound effects tied to a core feature, such as the audio quality of a headset.
• Paid AI recommendations, such as a chatbot answer a brand sponsored, which should say "Sponsored by [Brand]".

Suggested label wording is "Audio/Video created using AI" or "Audio/Video enhanced using AI". Platform labels can be used, and other wording is allowed if it is accurate. Labels must also follow ASCI's existing rules on disclaimers.

What does not need a label?

Routine editing is exempt: colour correction, lighting fixes, noise reduction and minor blemish removal. So are decorative backgrounds, ambient music, obviously fantastical effects, AI-written ad copy, and accessibility uses such as subtitles and translations. Exempt content still has to comply with the full ASCI Code.

What is banned even with a label?

The guidelines rule out fabricated endorsements or testimonials, AI visuals that show results a product cannot deliver, fake locations presented as real, unauthorised use of copyrighted work, and using someone's likeness without consent. As Saraf and Partners puts it, "disclosure does not cure an advertisement that is otherwise misleading". For beauty, wellness and fitness campaigns, that matters most: an AI-smoothed before-and-after is a results claim, not a cosmetic edit.

The draft, as MediaNama reported in June, also banned AI-generated fake authority figures, such as an invented doctor endorsing a supplement. The summaries of the final text we read do not repeat that example, but such an ad would almost certainly still fail as a fabricated endorsement.

How does this fit with creator disclosure rules?

These labels sit on top of ASCI's existing influencer rules, not in place of them. A paid creator post using a cloned voice or AI-generated product shots would need both the usual ad disclosure and an AI label. ASCI's message, as afaqs reports, is that "the responsibility remains with advertisers", so the brand cannot pass the duty to the creator or the AI tool.

Our take from Delhi

Most creator campaigns we see in Delhi NCR use AI quietly, often in edits the brand never reviews. The fix is cheap if it happens before December: ask creators to declare every AI tool used in a deliverable, and decide the label at the script stage. Brands should also stop treating AI demo shots as harmless filler, because that is exactly where ASCI will look first.