Keep Up IP law

by Rinmayee Khobragade

On 29 September 2026, the Advertising Standards Council of India (ASCI) released its Guidelines for Responsible Labelling of Synthetically Generated Content in Advertising. As generative tools become a routine part of how brands produce campaigns, the guidelines address a question advertisers have been asking for some time: when must the use of synthetic content be disclosed, and when is it not permitted at all? ASCI’s answer is anchored in the consumer rather than the technology. The guidelines will take effect three months from the date of publication.

The Regulatory Approach: Impact Over Technology.

The guidelines do not treat the use of AI as problematic in itself. Their concern is the effect that synthetically generated content (SGC) has on the person viewing the advertisement. SGC becomes misleading or harmful where it creates unrealistic expectations, exploits vulnerable groups, depicts unsafe situations, uses a person’s likeness without consent, or conveys false or misleading information. Every advertisement using SGC remains subject to the ASCI Code, which requires advertisements to be legal, decent, honest and truthful. Within that framework, the guidelines organise the analysis around four areas.

What remains prohibited. Advertisements that are illegal, infringe rights, make misleading claims or otherwise breach the ASCI Code remain prohibited, and attaching an AI label does not change that position. The guidelines identify fabricated endorsements and testimonials, synthetic representations that exaggerate product results or features in a misleading manner, non-existent locations or settings presented as real so as to mislead, the use of unauthorised copyrighted work, and the use of deepfakes or an individual’s likeness without consent.

The Prohibited Category. Labelling is required where SGC materially influences consumer decisions and the absence of disclosure could mislead. This covers synthetically generated influencers and brand ambassadors, the replication of a real person’s likeness or voice with consent for personalised messaging, fabricated events or settings that may shape consumers’ understanding of the product or service, demonstrations of products that do not yet exist, and synthetic sound effects that are central to a product’s core features, such as in advertising for headphones. Paid or sponsored AI-generated product suggestions attract a specific requirement: they must be labelled “Sponsored by [Brand]”.

The treatment of likeness deserves particular attention, since it sits across two categories. Without consent, the use of an individual’s likeness is prohibited outright. With consent, the replication of a likeness or voice for personalised messaging becomes permissible, but it must still be labelled. Consent therefore moves the use out of the prohibited category without relieving the advertiser of the duty to disclose.

Where no label is needed. Disclosure is not required where SGC has no material bearing on the consumer’s ability to make an informed choice. Routine editing such as colour correction, lighting adjustment, noise reduction and minor blemish removal falls within this category, as do decorative or ambient elements, obviously fantastical effects, administrative and text-based uses such as drafting advertising copy, and accessibility tools including subtitles, translations and accurate audio descriptions.

How disclosure should be made. Where a label is required, brands may use wording such as “Audio/Video created using AI” or “Audio/Video enhanced using AI”, and may rely on labels supplied by advertising platforms. Alternative wording is acceptable, provided it accurately conveys the nature of the synthetic content. Disclosures should also conform to the ASCI Code’s guidelines on disclaimers, wherever applicable. ASCI’s Secretary General and CEO, Manisha Kapoor, emphasised that an AI label cannot cure an advertisement that is otherwise misleading, and that responsibility for an honest and compliant end communication rests with the advertiser.

How the final text differs from the draft, the final guidelines follow a public consultation on a draft released in May 2026. The draft adopted an explicitly risk-based structure, classifying content as high risk (prohibited), medium risk (labelling required) or low risk (no labelling required), and was stated to be aligned with the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, 2026. The final version replaces the risk-tier terminology with the four areas described above, and the announcement accompanying it makes no reference to the IT Rules.

Two examples from the draft are also not expressly carried forward. The draft listed as prohibited the use of AI to generate fictional authority figures, such as a synthetic doctor promoting a supplement. It also required labelling for synthetic visuals depicting product performance unless those visuals replicated how the product actually performs. Neither example appears in the corresponding lists in the final announcement, although the broader prohibition on misleading claims and the materiality-based labelling requirement may well capture much of the same conduct.

Issues Awaiting Clarification, the guidelines take effect three months from publication, but the date from which that period runs is not separately specified, and it is not stated whether advertisements already in circulation will be covered. Nor does the announcement prescribe the placement, size or duration of AI labels beyond referring to the ASCI Code’s disclaimer guidelines.

What this means in practice.

For advertisers, the relevant enquiry is not whether AI was used but whether a synthetic element could materially affect how a consumer understands the advertisement or decides to buy. The underlying claim should be tested before any question of labelling arises, since no label will save a fabricated testimonial or a misleading depiction of results. Where a real person’s likeness or voice is to be replicated, consent should be secured and documented, and the output labelled. Sponsored AI-generated recommendations require the prescribed “Sponsored by [Brand]” wording rather than a generic AI label, and brands whose products have a sensory core feature, such as audio equipment, should look carefully at their use of synthetic sound. For agencies and platforms, the fact that responsibility rests with the advertiser makes it sensible for contracts to allocate clearly who decides on labelling. Care should also be taken that a label reflects what was actually done: content created using AI should not be described as merely enhanced.

Concluding thoughts. These are self-regulatory guidelines, to be read with the ASCI Code, against which every SGC-based advertisement will continue to be assessed in full. Their significance lies less in any single rule than in the principle they establish: transparency is required where it matters to the consumer, and it is never a substitute for honesty.