AI-Generated Ad Disclosure in India: A Practical Workflow for Brands

AI-generated advertising has moved from an experimental creative shortcut into a disclosure and governance problem. The practical question is no longer simply whether a team used AI. It is whether synthetic media could change what a reasonable customer believes about the product, person, event or evidence shown in the ad.

Google's July 2026 update added a global "How this ad was made" section to My Ad Center for ads across Search, YouTube and Discover. Google says its own generative advertising tools trigger an automatic disclosure, while advertisers will receive a control for declaring generative media created elsewhere. Read Google's announcement.

Meta updated its advertising-transparency approach in June 2026. It says "About this ad" will include AI information for media created or significantly edited with Meta's tools and that it is beginning to detect third-party AI creation or editing through industry-standard signals. Review Meta's current explanation.

For an Indian brand, those platform changes sit alongside India's intermediary rules and ASCI's advertising standards. They should be tracked separately rather than compressed into the vague instruction "label anything made with AI."

What changed for AI-generated ads in 2026?

Four developments matter.

The distinction matters. Platform labels are product behaviour. MeitY's rules are law directed principally at intermediaries. ASCI's Code is an established self-regulatory standard for advertisers, agencies and media, while the AI-specific risk framework reviewed for this article was still a draft. This article is an operating guide, not legal advice.

Do India's 2026 IT rules apply directly to every advertiser?

Not in the simple way many summaries imply.

MeitY's official FAQ says the 2026 amendments strengthen due diligence for intermediaries, including labelling, metadata or identifier duties for permissible synthetically generated information and additional obligations for significant social media intermediaries. The FAQ also says it is explanatory rather than a substitute for the notified rules. Read MeitY's official FAQ.

A D2C brand buying media is not automatically identical to an intermediary merely because it uses an AI image tool. The correct question is which entity creates, hosts, modifies, publishes or intermediates the synthetic content and which sector-specific rules apply. Legal review becomes especially important for health, finance, children, political communication, regulated products, impersonation or realistic synthetic people.

The useful operational conclusion is narrower: a brand should retain enough provenance and approval evidence to make an accurate declaration when a platform requests one and to support the ad if a regulator, platform or consumer challenges it.

Is ASCI's AI-advertising framework final?

The official document reviewed on 18 August 2026 is titled Draft Guidelines for Responsible Labelling of AI-Generated Content in Advertising. ASCI opened it for consultation through 13 June and said finalisation would begin afterward. HML did not find a final AI-specific version in ASCI's official press archive during this review.

The draft is still useful because it separates risk by likely consumer impact:

• high-risk material would remain prohibited even with a label, including fabricated endorsements, misleading product performance and unconsented likenesses;
• medium-risk material would require labelling when AI use materially affects a consumer's decision, such as realistic synthetic events or virtual ambassadors;
• low-risk uses would include minor corrections, decorative backgrounds and administrative text uses that do not alter the substance of the claim.

That is a sound creative-review lens, but the status must remain explicit: it is a proposal, not a final rule quoted as settled policy. Read ASCI's draft release.

When should an AI-assisted ad be escalated?

Use a materiality test rather than asking only whether AI touched the file.

Does the ad depict a real person or their voice?

Require documented consent, approved scope and a review of whether the synthetic output implies an endorsement the person did not make. A disclosure does not cure impersonation or an unauthorised testimonial.

Does the ad demonstrate product performance?

Compare the synthetic output with what a customer can actually experience. An AI-rendered stain disappearing, skin changing, food expanding or furniture fitting into a room can become a product claim even when no words make the claim.

Does the ad create a realistic event, customer or location?

Ask whether a reasonable viewer could mistake it for documentary evidence. Synthetic crowds, retail stores, customer reactions and before-and-after scenes deserve more scrutiny than an obviously fantastical backdrop.

Does the ad use a fictional expert or authority?

Do not use a synthetic doctor, engineer, analyst or customer to imply independent validation. The ASCI Code requires factual claims to be supportable and prohibits misleading representations by implication or omission. Review the ASCI Code.

Could the media affect a vulnerable audience?

Escalate creative involving children, health anxieties, financial insecurity, body image, safety or other situations where synthetic realism may exploit limited knowledge or heightened vulnerability.

What should an AI creative register contain?

A lightweight register should travel with every asset from generation to publication.

Do not rely only on file metadata. Social platforms, exports and messaging apps can change or strip metadata. Keep the register in the brand's own system and retain the approved source master.

How would this work for a 60-asset monthly pipeline?

Consider a hypothetical Indian skincare brand producing 60 paid-social assets per month:

• 20 use AI only for copy variants or minor background cleanup;
• 18 place real pack shots into synthetic lifestyle scenes;
• 12 use a synthetic person demonstrating the routine;
• 6 generate before-and-after product results;
• 4 use a founder's approved voice clone for language variants.

This is an HML operating example, not client data.

The team should not apply one disclosure decision to all 60 files. A workable triage would be:

1. Low materiality: copy assistance and minor cleanup enter the register and receive normal claim review.
2. Context materiality: synthetic lifestyle scenes are checked for false store, location, scale or usage implications.
3. Human representation: synthetic people and voice cloning require consent, likeness and disclosure review.
4. Performance representation: before-and-after creative requires the underlying result to be supportable; if the output exaggerates reality, a label does not make it acceptable.

The methodology is intentionally conservative. It ranks the risk that synthetic media changes a buyer's understanding, then adds the platform-specific declaration after the claim and rights review. It does not assume that an automatic platform label satisfies every legal or self-regulatory requirement.

The ROAS Calculator can help evaluate campaign economics after compliant assets launch. It cannot measure whether an AI disclosure was adequate or whether a claim was lawful.

What should the approval workflow look like?

Use seven gates.

1. Declare AI use internally. The creator records the tool, prompt or brief and material edits.
2. Classify materiality. Decide whether the output changes a person, claim, product demonstration, setting or evidence.
3. Verify rights. Confirm licences, model releases, voice permission and reference-asset rights.
4. Substantiate claims. Attach evidence for every factual or implied performance claim.
5. Choose disclosures. Complete Google or Meta controls and add an on-asset disclosure when the applicable review requires it.
6. Preview the live ad. Check the actual placement, transparency panel, landing page and accessible disclosure.
7. Archive and monitor. Keep the final asset, ad ID, approval and correction owner.

This workflow belongs beside the broader performance-marketing service and attribution glossary, because trustworthy creative and trustworthy measurement are both inputs to a defensible growth decision.

How should brands measure disclosed AI creative?

Do not frame disclosure as a conversion-rate trick. The first objective is accurate consumer information and reduced governance risk.

Performance analysis can still compare creative, but avoid a simplistic "label versus no label" test when the underlying media differs. Track:

• rejection or limited-delivery rates;
• correction and rework hours;
• complaints or trust signals;
• view-through and click-through behaviour;
• landing-page quality;
• conversion rate and contribution margin;
• post-purchase issues linked to misunderstood product representation.

For disciplined campaign interpretation, use HML's AI bidding guide and incrementality testing guide. Neither platform attribution nor AI optimisation can validate a deceptive claim.

The bespoke tailoring performance case study shows the wider operating context for coordinated Google and Meta execution; it is not evidence for the synthetic 60-asset example.

What should an Indian brand do next?

Audit the last 30 days of AI-assisted ads. Build an asset register, flag synthetic people and product-performance scenes, verify claim evidence, and record the platform disclosure state. Do not wait for an ad rejection to discover that nobody can reconstruct how a creative was made.

If your team produces AI-assisted creative at scale but lacks a review trail, request an AI creative governance sprint. The deliverable should be a risk matrix, asset-register template and channel approval checklist, not a promise that a label guarantees compliance.

Sources

1. Expanding AI transparency in ads - Google; published 9 July 2026; accessed 18 August 2026.
2. Expanding GenAI Transparency for Meta's Ads Products - Meta; updated 1 June 2026; accessed 18 August 2026.
3. FAQs on the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, 2026 - Ministry of Electronics and Information Technology; accessed 18 August 2026. MeitY states that the FAQ is explanatory and does not replace the notified rules.
4. Draft Guidelines for Responsible Labelling of AI-Generated Content in Advertising - Advertising Standards Council of India; published 12 May 2026; accessed 18 August 2026.
5. The ASCI Code - Advertising Standards Council of India; accessed 18 August 2026.