AI in Advertising: What India’s New ASCI Guidelines Mean for Brands and Businesses
The most common question raised in the commentary on ASCI’s new guidelines on AI is when should an ad be labeled as AI? A better question for brands is what a label cannot fix. The Guidelines were released on 29 September 2026 and will be effective from 29 December.
The short answer is that a label is disclosure, not a defence. Fabricated testimonials, exaggerated product results and deepfakes made without consent stay prohibited even when labelled. ASCI’s chief executive made the same point, adding that responsibility rests with the advertiser.
The more serious risk, however, is not in ASCI, but is self-regulatory. According to the 2022 guidelines issued by the CCPA, an endorsement should be based on the endorser’s genuine, reasonably current opinion, and penalties run up to ₹10 lakh for a first-time violations and ₹50 lakh for later ones. Courts are moving too: the Delhi High Court treated deepfake videos showing Ramdev endorsing products as prima facie misappropriation.
Consent is at the center but that opens one gate. The use of another person’s likeness is prohibited. However, even when a digital copy is consented for personalized messaging, it should be labelled. Consent and disclosure are two distinct boxes to tick and endorsement contracts should now include an AI-likeness clause that includes permitted uses, length of time, rights of approval and deletion on exit.
Beyond likeness, the most forward-looking provision is the easiest to miss. Paid AI product suggestions must be labelled “Sponsored by [Brand]”, and the Guidelines illustrate this with a chatbot recommending a moisturizer. Yet synthetic content is defined as audio, visual or audio-visual material, and the IT Rules category excludes AI-written text. This responsibility is perhaps unique. Such answers are live, so pre-release ad approval won’t be able to pick them up and brands will have to get labelling commitments from AI platforms in contracts.
A single asset may be subject to two tests. The IT Rules mandate that platforms label permitted synthetic content and for large platforms, that platforms gather declarations from users, while ASCI only mandates the labelling where the content has a material influence on the consumer. A conclusion without a label is thus not the answer to a platform’s declaration question.
Brands should test each AI asset against the three assessment questions in the Guidelines: Is it integral to the pitch? Can it mislead in performance? Would silence create a false impression? Before 29 December. Then apply the platform test above. Brands also need to document the basis for each labelling call, brief agencies and platform partners, and add likeness and labelling clauses to talent and AI-vendor contracts.
