The Advertising Standards Council of India has proposed a three-tier, risk-based framework for labelling AI-generated advertising, and one category matters most to creators: synthetic or "virtual" influencers, which would now require mandatory disclosure. It is a first attempt to draw a line between a real human creator and a manufactured one — but the loopholes are wide enough to matter.
What do ASCI's draft AI labelling guidelines actually say?
They sort synthetic advertising content into three risk tiers. As reported in trade coverage of ASCI's draft, released for public comment through mid-June 2026, "high risk" content — fabricated endorsements, fake testimonials, deepfakes and unauthorised likeness replication — is prohibited outright, disclosure or not. "Medium risk" content must be disclosed: AI-generated influencers, consented likeness replicas, and synthetic product-performance visuals that could sway a purchase. "Low risk" edits like blemish removal, colour correction or fantastical elements need no label.
Why does a virtual influencer need a disclosure label at all?
Because an audience deciding whether to trust a recommendation deserves to know whether a person actually made it. A virtual influencer is a wholly synthetic persona with no lived experience of the product it promotes, yet it is styled to read as a relatable human. India's influencer market — on track to cross ₹5,000 crore this financial year — increasingly features these AI avatars competing directly with human creators for the same brand budgets. A disclosure rule at least forces the synthetic option to identify itself. Why consent and transparency, not the technology, are the real battleground is the subject of our explainer on generative AI and rights.
What is the risk hiding inside the "medium risk" tier for real creators?
The category quietly legitimises replicating a real creator's likeness "with consent" — and consent is exactly where value leaks. A brand that secures a broad, one-time permission can generate an AI version of a human creator for campaigns they never shot, in languages they never spoke, indefinitely. If that reuse is not separately itemised and priced, the creator has handed over a reusable asset for the price of a single shoot. Our likeness licensing rate benchmarks for India exist precisely so creators can put a defensible number on each AI-usage right rather than bundling it away.
Will these guidelines actually be followed?
That is the honest worry. ASCI is a self-regulator that acts reactively, and commentators have noted that roughly three-quarters of top Indian influencers already flout basic paid-partnership disclosure — so a new layer of AI labelling, built on subjective terms like "minor enhancement," may be widely ignored. Monitoring synthetic content at India's scale is close to impossible manually. The practical lesson for creators is not to wait for enforcement to protect them, but to build their own record of what they did and did not permit. The vocabulary of these permissions is spelled out in our rights and AI glossary.
What should Indian creators and brands do now?
Treat AI usage as a named, separately-priced permission in every contract, not a throwaway clause. Creators should register a timestamped assertion of their likeness and voice, state which synthetic uses are allowed and at what rate, and keep that record enforceable rather than buried in email. Brands, for their part, gain from clean consent too: an itemised, auditable permission is what keeps a campaign on the right side of both ASCI's guidelines and personality-rights case law. That register-set-price-license workflow is the whole point of how the Zimorta model works.
The bottom line
ASCI's draft is a useful signal that Indian advertising is starting to distinguish the synthetic from the real. But labels alone will not protect creators whose likeness can be replicated "with consent." The creators who stay ahead will define that consent themselves — precisely, priced and on the record — before a brand or an avatar defines it for them.
