ASCI AI advertising labelling guidelines: what to label from December 2026

Summary. The Advertising Standards Council of India (ASCI) released its Guidelines for Responsible Labelling of Synthetically Generated Content in Advertising in Mumbai on 29 September 2026, after a May draft and a consultation that closed on 13 June. The guidelines start when three months have passed from publication, and ASCI has printed no calendar date. Counting from the 29 September release gives 29 December 2026; the signed document is dated 17 September 2026, which gives 17 December. We recommend having live creative compliant by 17 December. The rules sort AI use into three groups. Prohibited even with a label: fabricated testimonials, exaggerated AI results, non-existent places shown as real, unauthorised copyrighted work, and deepfakes or likenesses used without consent. Label required: AI influencers, consented voice or likeness clones in personalised messages, fabricated settings that affect how a product is understood, demos of products that do not exist yet, product-critical sound effects, and paid AI product suggestions, which must read Sponsored by [Brand]. No label: routine edits, decorative elements, obvious fantasy, AI-written copy and accessibility uses. Separately, a MeitY advisory dated 8 October 2026, published by MediaNama but not by the ministry, asks social media platforms to label or contextualise manipulated, synthetic or misleading content. This is general information, not legal advice.

About this article. For marketing heads, agencies and in-house legal teams running AI-assisted advertising in India. On 11 November 2026 we checked the final ASCI guidelines and May 2026 draft, ASCI's press releases, MeitY's consolidated IT Rules (updated as on 10 February 2026) and Meta's Business Help Centre. MeitY's 8 October advisory is not on the ministry's website, so we use the text MediaNama reproduced and label it as reported. Not legal advice; no vendor paid for inclusion. Researched with AI assistance and reviewed by our editorial team.

When the guidelines take effect: 17 or 29 December

The guideline document ends with a rule and no date: the guidelines come into effect when three months have passed from their publication. It is signed for ASCI's Board of Governors by its Chairman and dated 17 September 2026. ASCI's press release, datelined Mumbai, 29 September 2026, says ASCI released the guidelines that day and repeats the three-month rule. ASCI's press announcements page lists the release under September 2026 and, as of 11 November, shows no later notice, FAQ or calendar date for it.

The 29 December date is our calculation from the release date, not an ASCI statement. Neither reading matters if your live creative is compliant by 17 December.

One point takes most of the pressure off the date. The prohibited uses are not new rules. The guideline document says synthetic content becomes misleading or harmful when it breaches Chapters I and III of the existing ASCI Code, and that every ad with synthetic content must comply with the whole Code. A fabricated AI testimonial already breaches the Code today. The three-month window mainly governs the new labelling duty.

ASCI is a voluntary self-regulatory body, not a government regulator. MediaNama notes that its decisions carry no legal penalty of their own, but that most of the uses it bans already fall under the consumer protection authority's 2022 guidelines on misleading advertisements, the Copyright Act or court rulings on personality rights.

What changed from the May 2026 draft

ASCI published draft guidelines on 12 May 2026 (the draft document itself is dated 8 May), said they were aligned with the IT Rules amendments of 10 February 2026, and took comments until 13 June. The comparison below sets the draft document against the final text.

Law firm Fox Mandal's summary notes two of these changes: the dropped AI doctor example and the dropped word exaggerated. For agencies, the most useful change is the sharper three-question test, which lets a reviewer make and record a decision rather than hunt for a matching example.

What Indian brands must label: the three groups

ASCI's assessment questions are, in substance:

1. Is the synthetic content central to the ad's persuasive message, or does it materially influence purchase decisions?

1. Left unlabelled, would the way it is used mislead consumers about the product's claims, performance or benefits?

1. Would the absence of a disclosure create a false or misleading impression of the product's claims, benefits or performance?

A yes pushes the asset towards a label. If the content would mislead even with a label, it belongs in the prohibited group.

Prohibited, even with a label

• Fabricated endorsements or testimonials. Using AI to create, alter or attribute an endorsement that falsely shows a real person used or recommended a product.

• Exaggerated results. ASCI's example is a weight-loss ad using an AI image of a toned body to suggest a drop from 100 kg to 70 kg that substantiated results do not support.

• Non-existent locations shown as real, such as a travel ad showing a hotel or facilities that do not exist.

• Unauthorised copyrighted work, deepfakes, or a person's likeness used without consent.

Label required

• Synthetically generated influencers and brand ambassadors.

• A real person's likeness or voice replicated with consent for personalised messaging.

• Fabricated events, settings or situations that may affect consumers' understanding of the product's performance.

• Demonstrations of products that do not yet exist, such as a 3D model of an unbuilt housing complex.

• Synthetic sound effects highly relevant to the product's core features, such as audio quality in a headset ad.

• Paid or sponsored AI product suggestions, labelled Sponsored by [Brand]. ASCI's illustration is a chatbot recommending a moisturiser for Mumbai weather because the brand paid for it.

No label needed

• Routine edits: colour correction, light adjustment, noise reduction, blemish removal, formatting and copy refinement that leave the substance and claims unchanged.

• Purely decorative backgrounds, abstract skylines, ambient music, jingles and background sounds unrelated to the product's capabilities.

• Obvious fantasy, such as a fairy fixing a product with magic dust or a car turning into a spaceship.

• AI-generated or AI-polished ad copy, audio descriptions and good-faith administrative documents.

• Accessibility uses: subtitles, closed captions, translations and accurate audio descriptions.

Quick classifier for common Indian ad formats

These rows are our reading of ASCI's examples applied to typical cases, not ASCI rulings. Edge cases still need judgement, ideally from your legal or compliance team.

Labels: wording, placement and what platforms add

ASCI does not prescribe one label. Brands may use labels provided by ad platforms, their own wording such as Audio/Video created using AI or Audio/Video enhanced using AI, or any other label that accurately tells consumers what is synthetic. Sponsored AI recommendations must state Sponsored by [Brand]. Disclosures should follow the ASCI Code's existing guidance on disclaimers, and ASCI states that a label does not rescue an ad that misleads anyway.

ASCI sets no size or duration for labels. MediaNama reports that MeitY proposed in April 2026 that labels on synthetic video stay continuously visible for the whole video. The consolidated IT Rules on MeitY's website are still the version updated to 10 February 2026, which asks for a prominent, easily noticeable label without a duration, so treat the continuous-label idea as a pending proposal.

Format by format. Our recommendations: for video, show a legible label while the synthetic element is on screen, not only on the end card. For audio-only ads, use a spoken disclosure. For static and display ads, put text on the creative, not only on the landing page. For influencer and AI-ambassador posts, label on screen and in the caption, alongside the paid-partnership disclosure. For personalised messages with cloned voices or faces, label every message.

What Meta adds on its own. Meta's Business Help Centre says Meta automatically adds an AI info label when an ad is created or significantly edited with some of its generative tools, such as Background Generation, Image Generation or Add Animation, and when it detects third-party AI editing through industry metadata such as C2PA. In most cases that label sits behind the ad's three-dot menu, under About this ad. Meta lists India among the regions (with the European region, California, New York and Taiwan) where advertisers can self-disclose generative AI use so that a label appears next to the ad's sponsored label.

ASCI allows platform labels, but it does not say whether a label a platform applies by itself, tucked inside a menu, meets its requirement, a gap MediaNama also flagged. Our working view: for anything in the mandatory group, put your own visible label in the creative and use the platform's self-disclosure as well. Meta said on 6 October 2026 that it is expanding its generative AI video tools for advertisers, which makes this a routine question for D2C teams; our guide to Meta's Q4 2026 ad changes covers those tools.

How ASCI, the IT Rules, MeitY's advisory and platform labels fit together

The table draws on ASCI's guideline document, MeitY's consolidated rules, MediaNama's reproduction of the advisory and Meta's Help Centre.

The IT Rules. MeitY amended the IT Rules through G.S.R. 120(E) dated 10 February 2026, and the changes took effect on 20 February. Under rule 3(3), a platform whose tools can create synthetic content must label it prominently, embed permanent metadata or another provenance marker where technically feasible, and not let anyone strip either. Under rule 4(1A), a significant social media intermediary must ask users before upload whether content is synthetic, verify the declaration and label what is confirmed. MediaNama puts that threshold at more than 50 lakh users in India, and notes that the Rules ask whether content is synthetic while ASCI asks whether it could mislead a buyer. The Rules also do not say whether platforms' duties cover ads bought through ad managers. Our article on the Gemini 2.5 Flash Image shutdown covers keeping SynthID and C2PA markers intact through an image pipeline.

MeitY's 8 October advisory (reported). MediaNama reproduced what it says is the advisory as issued: File No. 16(2)/2026-CyberLaws, dated 8 October 2026, from MeitY's Cyber Laws Division, signed by Joint Secretary Ajit Kumar and addressed to all social media intermediaries. Its clause (iii) asks platforms to identify and label or contextualise manipulated, synthetically generated or misleading content, particularly content presented as a current event. It also flags old footage shown as current, synthetic audio-visual content and impersonation of public authorities or persons, applies with immediate effect and lists no penalties. SFLC.in says the advisory has not been made public and describes it as non-binding. MediaNama points out that its labelling clause, unlike the February Rules, is not limited to audio-visual content.

For advertisers the advisory is indirect, because it is addressed to platforms. Its practical effect is on topical moment marketing: a synthetic news-style clip, a fake breaking announcement or an AI likeness of an official in an ad is more likely to draw a platform label or removal. If it misleads or uses a likeness without consent, it is also prohibited under ASCI's guidelines.

Workflow changes for brands and agencies

AI use now has to be recorded per asset. A workable process:

Three practical details matter more than the table suggests.

Re-check always-on and evergreen creative. The guidelines contain no exemption for ads launched before the effective date, so treat anything still running after it as in scope. Catalogue videos, AI ambassadors and personalised voice campaigns launched in 2026 need the same classification before 17 December.

Write it into contracts. Agency and production contracts should state who classifies, labels and holds consent records, and require suppliers and influencers to disclose AI use. Ask your counsel to word the warranties.

Keep provenance metadata. Meta says it uses C2PA metadata to detect third-party AI edits. Stripping metadata to avoid a platform label is a poor trade: it weakens your audit trail and does not change whether ASCI requires a label.

Why records matter: ASCI's homepage showed 91,665 complaints handled to date when we checked on 11 November 2026. If a complaint arrives, a dated classification with a named reviewer is the quickest way to show the decision was considered.

A timeline to 17 December

Worked example: one D2C campaign through the new process

Take an illustrative skincare brand planning a year-end campaign with six AI-assisted assets.

1. A virtual brand ambassador in a 20-second reel: label required. Add Video created using AI on screen and in the caption.

1. The founder's voice cloned, with written consent, for personalised WhatsApp offers: label required, plus the consent record.

1. An AI before-and-after image showing results beyond what the brand's trial data supports: prohibited. Replace it with substantiated imagery.

1. An AI beach background behind a product shot in a static ad: no label, provided it is decorative and implies no real location claim.

1. Captions translated into Hindi and Marathi by an AI tool: no label; check the translation.

1. A paid placement in an AI shopping assistant: label required, naming the brand as sponsor.

Result: three labels (assets 1, 2 and 6), one removal (asset 3) and two assets that need nothing (4 and 5). The removal is the expensive one, and it breached the Code before December too.

Asset 6 has a gap: ASCI does not say how an advertiser labels a suggestion the AI system itself writes. MediaNama notes that OpenAI says its ads in India appear separately from answers and carry a label. Check whether the platform's label names your brand. Our guide to ChatGPT ads in India explains how those ads are bought, and our AI Mode ads playbook covers Google's Sponsored label on ads shown with AI answers.

How eCorpIT can help

eCorpIT is a Gurugram-based technology organisation, founded in 2021, assessed at CMMI Level 5 and MSME certified, with senior-led engineering teams working across AWS, Microsoft and Google platforms. Our performance marketing team can inventory AI-assisted creative across your ad accounts, build declaration and classification steps into briefing and approval, and keep an audit trail for each asset. Talk to us at /contact-us/.

Last updated: 11 November 2026.

References

1. ASCI Releases Guidelines for Responsible Labelling of Synthetically Generated Content in Advertising, press release, 29 September 2026 — ASCI

1. ASCI Releases Draft Guidelines for Responsible Labelling of AI-Generated Content in Advertising, press release, 12 May 2026 — ASCI

1. Guidelines for Responsible Labelling of Synthetically Generated Content in Advertising, signed 17 September 2026 — ASCI

1. Exclusive: Amid protests, MeitY widens labelling for social media beyond synthetic content, 8 October 2026 — MediaNama

1. Press Announcements — ASCI

1. The ASCI Code — ASCI

1. India's new AI ad labelling guidelines: What brands must disclose and when AI content counts as an ad, 6 October 2026 — MediaNama

1. Draft Guidelines for Responsible Labelling of Synthetically Generated Content in Advertising, dated 8 May 2026 — ASCI

1. ASCI Issues Guidelines for Labelling of AI Content in Advertising, 29 September 2026 — Fox Mandal

1. The Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, updated as on 10 February 2026 — Ministry of Electronics and Information Technology

1. About AI info on ads created or edited with generative AI tools — Meta Business Help Center

1. Advertising Week New York 2026: New AI Capabilities to Guide Campaigns and Reach Customers, 6 October 2026 — Meta for Business

1. ERGO: IT Rules Amendment, 16 February 2026 — Khaitan & Co

1. Statement on MeitY's Advisory to Social Media Intermediaries regarding prevention of dissemination of misinformation, 10 October 2026 — SFLC.in

1. Home page, complaints handled till date counter — ASCI