ASCI’s AI labelling rules: what hospital marketing should change before the year ends
On 29 September, the Advertising Standards Council of India issued guidelines on labelling AI-generated content in ads. They do not mention healthcare once. Hospital marketing teams should read them anyway, because AI doctor videos, cloned voices and rendered facilities are already part of how hospitals advertise.
Key facts at a glance
| What happened | ASCI released its Guidelines for Responsible Labelling of Synthetically Generated Content in Advertising on 29 September 2026. They take effect three months after publication. |
| Reported by | ASCI press release, 29 September 2026 |
| My take in one line | For hospitals, an AI label is not a disclaimer. It is a trust signal, and the teams that adopt it early will look more credible, not less. |
What happened
ASCI’s new guidelines say an AI label is required when synthetic content could materially change how a consumer understands an ad. The listed cases include synthetic influencers or brand ambassadors, replicating a real person’s likeness or voice (with consent), fabricated events or settings that affect how a product is understood, demonstrations of products that do not exist, and sponsored product suggestions inside AI tools, which must carry a “Sponsored by” label.
Routine editing such as colour correction, decorative elements, obvious fantasy effects, AI-written ad copy and accessibility uses like subtitles and translation do not need a label. Some uses stay off-limits with or without a label: fabricated endorsements or testimonials, exaggerated results, presenting non-existent places as real, and using someone’s likeness without consent. Suggested label wording includes “Video created using AI” and “Video enhanced using AI”.
My take
Healthcare is not named, but hospitals are among the heaviest users of exactly the formats the guidelines cover. A senior consultant’s face and voice, cloned to deliver an awareness video in six languages. A walkthrough of a hospital that is still under construction. A patient story where the patient is an AI actor because consent was hard to get.
Each of these now has a clear rule. The first needs consent and a label. The second needs a label if a viewer could mistake the render for the real building. The third is not a labelling question at all. A synthetic patient testimonial is a fabricated testimonial, and the guidelines do not allow it.
This is a good development for hospitals, not a burden. Patients already judge hospital marketing more sceptically than most categories. A visible, honest label tells them the hospital has nothing to hide about how its content was made. In a category built on trust, that is worth more than the few seconds of polish the label might cost.
What most coverage missed
The clause on sponsored AI suggestions matters more for hospital growth teams than the deepfake rules. As AI assistants start answering “which hospital should I go to for a knee replacement”, any paid placement in those answers will have to be labelled as sponsored. That keeps a clear line between earning a citation and paying for a mention. Hospitals that have invested in content AI systems can trust and cite will hold the unlabelled, organic position. That is the core argument for generative engine optimisation in healthcare.
The second point is that these rules stack on top of what hospitals already follow. NMC norms on doctors’ advertising and the Drugs and Magic Remedies Act still apply. An AI label does not make a non-compliant claim compliant.
What I would do
- Build a synthetic content register. List every live ad, video and landing page that uses an AI avatar, cloned voice, rendered facility or generated imagery, with the label status of each.
- Get written consent from every doctor whose face or voice is replicated. Define scope, languages, channels and an expiry date, and keep it with the asset.
- Label pre-opening renders. For a hospital launch, a simple “Artist’s impression, created using AI” line protects the brand when the real building opens.
- Rule out synthetic patient stories completely. If consent for a real story is not available, use a doctor explaining the condition instead.
- Add one line to every agency brief and approval checklist: does this asset need an AI label, and where does it sit?
- Keep using AI for subtitles, translation and copy drafts. These are exempt, and they are where most of the efficiency sits anyway.
What to watch
- The first complaints ASCI processes under the new guidelines once they take effect, and whether any involve health brands.
- Whether platform-applied AI labels on Meta and YouTube are treated as enough, or brands are expected to add their own.
- How AI assistants in India begin to show sponsored health answers, and how clearly they label them.
Source: ASCI, Guidelines for Responsible Labelling of Synthetically Generated Content in Advertising. Figures as reported at the time of writing.
Questions people ask
They are rules ASCI released on 29 September 2026 that require ads to disclose AI-generated or AI-altered content when it could materially affect how a consumer understands the ad. They take effect three months after publication.
Yes. They apply to all advertisers. Healthcare is not singled out, but hospital ads that use AI doctor avatars, cloned voices, rendered facilities or synthetic people fall within the cases that need a label.
Only with the doctor’s consent, and the ad must be labelled as created or enhanced using AI. NMC norms on doctors and advertising continue to apply to what the doctor is shown saying.
No. ASCI’s guidelines prohibit fabricated endorsements and testimonials whether or not they are labelled.
No. Accessibility uses such as subtitles, translations and audio descriptions are listed as exempt, as are routine edits and AI-written ad copy.
Create a register of all AI-assisted assets, collect written consent from doctors whose likeness or voice is replicated, label pre-opening renders, remove any synthetic patient stories and add an AI label check to the approval workflow.
Read my takes first in Google Search
