India Synthetic Media Advertising Rules 2026: The ASCI 3-Tier Framework for Commercial Video

India synthetic media advertising rules 2026: the two instruments

India synthetic media advertising rules 2026 give commercial video teams a clear decision tree for any AI-generated ad: ban it, label it, or leave it alone. The Advertising Standards Council of India's three-tier framework and the Ministry of Electronics and Information Technology's SGI amendments together decide what must be disclosed before a cut goes live.

Two separate instruments now govern synthetic media in Indian advertising. The MeitY amendment to the IT Rules, notified on 10 February 2026 and live from 20 February, sets the statutory backbone: it defines synthetically generated information and forces platforms to label and trace it. ASCI's draft guidelines, released in May 2026, translate that backbone into a practical advertiser checklist.

The split matters because the two layers punish different failures. MeitY reaches the platforms that host and generate content; ASCI reaches the brands that pay for and publish the ad. A video team therefore has to satisfy both at once: build the asset so the platform can label and trace it, and brief the creative so the message lands in the right tier. Treating either layer as someone else's problem is exactly how finished cuts get pulled after launch.

The ASCI three-tier framework: banned, labelled, exempt

ASCI sorts every AI-affected ad into three risk buckets by consumer outcome, not by the tool used. High-risk content is prohibited outright - fabricated endorsements, exaggerated product claims, fake but realistic locations, unauthorised deepfakes, and fictional authority figures such as a synthetic doctor pushing a supplement. A label cannot rescue content that breaks the ASCI Code, because the harm is in the claim, not the disclosure.

Medium-risk content must carry a disclosure. This covers virtual influencers, cloned voices, synthetic product demonstrations, AI-built scenes, and products that do not yet exist, such as a 3D model of an unbuilt tower. Teams already juggling multiple regimes can fold India into the same {{link}} they use for the EU, New York and China. Low-risk use - colour correction, ambient sound, fantasy creatures - needs no label at all, because no reasonable viewer mistakes it for reality.

The clearest signal in the framework is that the label is a floor, not a ceiling. ASCI is explicit that correctly labelled content can still be judged misleading if the end effect deceives a viewer, so disclosure is the minimum the brand owes, not a shield against scrutiny. For production teams this means the tier assignment should happen at storyboard lock, not in legal review after the cut is rendered. A synthetic spokesperson, a cloned voice and a generated product shot each need their disclosure decision made before a single frame is generated.

Teams already juggling multiple regimes can fold India into the same cross-market disclosure map they use for the EU, New York and China.

MeitY's SGI rules: labelling and provenance at the platform layer

The IT Rules amendment defines SGI as audio, visual or audiovisual content created or altered by a computer resource so it appears real and indistinguishable from a genuine person or event. Pure text is excluded, and routine good-faith editing such as noise reduction or compression is carved out. Permitted synthetic content must carry a prominent label and permanent provenance metadata that survives the export.

Platforms enabling synthetic generation must deploy technical measures to block unlawful SGI and embed a unique identifier that traces the content back to the tool that made it. The obligation to embed permanent metadata lines up with the C2PA-based {{link}} buyers now expect from commercial cuts. Takedown timelines were also cut from 36 hours to three, with stricter clocks for intimate or impersonation material, so a flagged asset has to move fast through the pipeline.

For a brand, the practical consequence is that provenance is no longer a nice-to-have buried in the spec sheet. The unique identifier has to survive the export, the edit and the platform hand-off, which means the generation tool, the editor and the distribution team all have to agree on one metadata convention. If a cut loses its marker in transit, the hosting platform is on the hook for letting unlabelled SGI through, and that liability flows straight back to the advertiser whose asset it is.

The obligation to embed permanent metadata lines up with the C2PA-based provenance record buyers now expect from commercial cuts.

How the two layers map onto your AI video production

For a commercial video team, the practical question is which production choices trip a tier. A synthetic brand ambassador or a cloned voiceover lands in medium-risk and needs a visible 'created using AI' label. A product-performance clip must show the product as it truly performs, or the enhancement itself becomes the claim. Fabricating a testimonial or a real person's likeness without consent is simply off the table.

Google's {{link}} already decides when a generated ad must carry a visible label; India's tiers reuse that instinct at the content level. The cleanest workflow is to score each shot against the three tiers during storyboard lock, attach the label in edit, and write the provenance marker at export so the asset arrives compliant rather than patched after launch.

A useful habit is to attach a tier tag to every shot in the production board, the same way you already tag aspect ratio or codec today. High-risk shots get killed at the pitch stage; medium-risk shots get a label column and a provenance column; low-risk shots get a written note explaining why no label is needed, which is itself a record if the call is ever challenged. The board becomes the audit trail, and the audit trail is what both MeitY and ASCI ultimately want to see.

Google's primary-creation test already decides when a generated ad must carry a visible label; India's tiers reuse that instinct at the content level.

Where India sits next to other markets

India is not alone in writing synthetic-media rules, but its risk-tier model is distinctive. Unlike {{link}}, which forces a visible 'AI-generated' mark on every synthetic ad, India scores risk by tier. Saudi Arabia's {{link}} builds consent and watermarking into the pipeline; India reaches the same outcome through platform-level SGI duties. The net effect for global brands is a fifth regime to clear on one asset.

The takeaway for multi-market campaigns is to treat disclosure as an asset-level property, not a campaign-level footnote. A single master cut can satisfy several regimes if the label, the provenance marker and the consent chain travel with the file from the first generation rather than being retrofitted per market after the fact.

The common thread across these regimes is that disclosure is moving from the campaign form into the file itself. What India adds is a graded model: rather than one blanket mark for everything synthetic, it asks whether the AI use actually changes what the consumer decides. That nuance is harder to automate, but it spares brands from labelling decorative backgrounds and fantasy creatures that no viewer mistakes for reality, which is exactly the label fatigue the rules were written to avoid.

Unlike South Korea's mandate, which forces a visible 'AI-generated' mark on every synthetic ad, India scores risk by tier.

Saudi Arabia's Gulf guidance builds consent and watermarking into the pipeline; India reaches the same outcome through platform-level SGI duties.

A pre-launch compliance checklist for India-bound video

Before any India-bound cut ships, run a three-point check. First, classify every AI-affected shot against the ASCI tiers and ban anything high-risk. Second, attach the correct label - 'Audio/Video created using AI' or 'enhanced using AI' - wherever medium-risk use appears. Third, confirm the platform hosting the asset will carry the provenance marker and honour the three-hour takedown clock.

None of this is optional, and a label is a floor rather than a ceiling: ASCI is explicit that correctly labelled content can still be pulled if the end effect misleads. Build the check into the brief, not the review, and the 2026 rules become a production constraint you design around instead of a takedown you fear.

The teams that will cope best are the ones that already run a generation gate. If a cut is scored, labelled and traced at export, India-bound delivery is a settings change, not a rework. The 2026 rules reward teams that treated provenance as infrastructure months ago, and they penalise teams that treated AI video as a fire-and-forget feed. Compliance, in other words, is mostly a workflow decision made long before the ad ever reaches a viewer.