AI Generated Synthetic Content Rules In India : What Brands Must Now Do

Introduction : On 10 February 2026, the Ministry of Electronics and Information Technology (MeitY) announced the Information Technology (Intermediates Guidelines and Digital Media Ethics Code) Amendment Rules, 2026, which introduce a new category - ‘Synthetically Generated Information’ (SGI), effective from 20 February 2026. No law in India, for the first time, explicitly classifies AI generated and AI altered AVC as a separate compliance category that imposes unique due diligence requirements on intermediaries. For the first time in India, AI generated and AI altered AVC will be recognised as a separate category of compliance and intermediary obligations will be imposed under the law. The Advertising Standards Council of India (ASCI) followed suit on 8 May 2026 with the issuance of draft guidelines for the responsible labelling of advertising content created by Artificial Intelligence (AI).

The process is far from peripheral for brands, advertising agencies and in-house marketing and legal teams. From copywriting to creating images, from synthetic voiceovers to virtual influencers, and from AI-generated product demonstrations to every other type of AI-enhanced content, generative AI is now woven into content pipelines. The 2026 regime changes the way in which content is judged for moderation and editorial decisions made by the platform, and means that there will be penalties for non-compliance, including takedowns at the platform level, loss of safe harbour protections and penalties from the ASCI. This blog explores the Indian legal framework when it comes to the regulation of synthetic content, its impact on creators, platforms and advertisers, and provides a compliance roadmap for marketing teams.

Legal Provisions

The IT Rules Amendment, 2026

According to Rule 2(1)(wa) of the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, amended which defines Synthetically Generated Information as information, whether in the form of audio, video or audio-visual content, that has been artificially or algorithmically created, generated, modified or altered, so that it appears to be authentic or true or appears to depict a real person, event or circumstance. As per Rule 3(3), content must be clearly and visibly labeled or embedded with metadata that indicates it is SGI at the time of upload and intermediaries are forbidden from removing, hiding or altering the label. Rule 4(1A) extends to Significant Social Media Intermediaries (SSMIs) having more than 5 crore users in India and mandates an affirmative pre-publication declaration by the uploader of whether the content is generated synthetically or not with independent automated verification by the platform. The amendments also narrow the time window for takedown actions against the most detrimental types of SGI, such as deepfakes and impersonation, to within 2-3 hours of including it in the list of "knowledge.

The amendments also introduce a time limit for takedowns of the most harmful categories of SGI, including deepfakes or impersonation content, of no more than two to three hours from knowledge, after which the intermediary may lose the safe harbour protection provided by Section 79 of the Information Technology Act, 2000.

Asci's Draft Guidelines On AI Generated Advertising

ASCI's draft guidelines allow for the use of AI, with the caveat that it must be clearly disclosed and that the three levels of risk are categorized as follows: low-risk (routine use of AI to edit the ad, colour correct it or add clearly fantastical elements, which do not require any disclosure); medium-risk (synthetic influencers, visuals of product performance generated using AI, replicating voices or likenesses, or recommendations from AI, which must be clearly disclosed, such as ‘Audio/Video created using AI’ or ‘Audio/Video enhanced using AI’); and high-risk (content that created unrealistic expectations, exploited vulnerable audiences, depicted unsafe conduct, or used a real person's likeness without their consent, which is banned for all intents and purposes, including labelling). Most importantly, ASCI points out that AI-driven advertisements of a celebrity's voice or appearance are still medium-risk and are required to be disclosed, even if that celebrity has agreed to the creation of their likeness by AI.

Legal Analysis 

Disclosure And Labelling As A Layred Obligation

The 2026 framework is designed on two different layers, both of which overlap. The IT Rules are platform facing, in that they create a duty for the intermediaries to affix and maintain labelling of SGI, and a duty to gatekeep uploads for SSMIs via mandatory declarations. The ASCI guidelines contain an added requirement that extends beyond the technical labelling of a video as AI-generated: If the AI usage underpinning an advertisement is misleading, even if the platform has correctly labelled the video as AI-generated, the advertisement itself may be in violation of the ASCI Code, regardless of whether a label is present. Therefore, relying on platform level compliance is not a basis for brands to rely on advertisement level compliance; each of these obligations must be discharged separately.

Provenance And The Metadata Problem

Rule 3(3) stipulates embedding of tamper-evident metadata that can be used to trace changes in the content’s origin, authorship, and history, with inspiration from nascent content-provenance standards like C2PA. In practical terms this can be a true challenge for marketing teams, as the provenance metadata is often lost through normal compression, cropping, re-encoding, or cross-platform re-sharing of an asset, so that what may have been a properly labelled asset may turn up at the point of publication without any provenance marker whatsoever. The legal and marketing teams should not assume that just one embedding will be enough to remove the obligation - the provenance should be checked and if required, re-affixed throughout the chain of distribution, such as when the creative is handed over to a media-buying agency or influencer partner for further editing.

Internal Approval Workflows

Under Rule 4(1A) the point of compliance is now the time of upload, meaning that brands will no longer be able to rely on post-publication review. Content approval workflows must have an explicit AI-use point before creative sign-off – It is essential that all content that goes into the production pipeline be identified (at the point of intake) whether or not AI was used (as a script, visuals, voice or in performance simulation) and what tier of ASCI risk that use introduces, and what platform declaration it entails – before content is briefed for output. This contrasts with the conventional vetting, which is only carried out at the final approval stage. SGI compliance involves tracking back from the first prompt.

Relevant Case Laws And Precedent

The compressed takedown windows and the requirement for SSMI verification are both new, and the constitutionality of the amendment Rules is widely expected to be challenged before the constitutional courts on Article 19(1)(a) basis. The analysis is, however, based on precedent. While the 2026 amendments arguably deviate from the doctrine of ‘constructive knowledge' or ‘actual knowledge' provided by the erstwhile Section 79 read with IT Rules, in Shreya Singhal v. Union of India (2015) 5 SCC 1, the Supreme Court interpreted intermediary liability as requiring the actual or constructive knowledge of the existence of material on their servers through a court or government order before the removal obligations arose. In Google India Pvt. Ltd. v. Visaka Industries, (2020) 4 SCC 162, the Supreme Court had held that the safe harbour of Section 79 of the IT Act is subject to a condition, namely the condition of compliance of the due diligence requirements set out in the IT Act.

Practical Implications And Risks For Marketing Teams

So, the regulatory change brings some tangible risks for brands and their agencies. The first, contractual risk: brands regularly purchase content from third party agencies, freelance producers and influencer partners who are using generative AI without notification to the brand; contractual obligations impose a warranty and/or indemnity expressly to cover the use of SGI by third parties – otherwise, the brand as the ultimate responsible party for the advertisement under the ASCI Code will be exposed to compliance risk. Second, platform risk: If content is uploaded incorrectly or is not uploaded, it may be refused outright by the SSMIs under Rule 4(1A), which can throw off campaign schedules, especially when the platform has a set date for the push of paid media. Third, reputational and regulatory risk: if the brand is endorsed by an undisclosed synthetic celebrity, or if a synthetic demonstration of the product's performance is made, it exposes the brand to an ASCI complaint, a reference for misleading advertising to the Central Consumer Protection Authority under the Consumer Protection Act, 2019 and to platform level takedown action.

Fourth, safe harbour risk for brand owned platforms: any brand that operates its own app, community forum or user-generated content feature (such as contests, testimonials, reviews) is now an intermediary with respect to such feature, and therefore will be required to meet their own labelling and metadata obligations, or risk losing safe harbour protection for user-uploaded content on their own platform.

Ultimately, another potential hidden hazard is the over-labeling or under-labeling due to uncertainty. The ASCI use tiers are fact-based and materiality-based determinations of whether a use of AI is likely to 'materially influence' a purchase decision, meaning that without documented internal criteria, marketing teams can make inconsistent labelling decisions between campaigns, leading to regulatory interest and issues should a complaint arise - and defence should it.

Comliance Checklist For Marketing Teams

• Identify at what stage and in which content pipeline (in-house, agency, freelancer, influencer) the generative AI tools are being used.

• Add a form for required AI use at project briefing, prior to the start of creative production.

• For each asset, categorize in ASCI's three-tier risk classification (low, medium, high) and record the reasons for the classification.

• Before publishing, put prominent disclosure labels (such as ‘Audio/Video created using AI’) on all medium-risk content, in line with ASCI guidelines.

• Deny or re-design any idea or concept that poses a high risk (unrealistic claims, use of likeness without consent, exploitation of vulnerable groups) regardless of any proposed labelling.

• Maintain or re-embed metadata regarding the origin of the information at each stage of editing, compression and/or cross-platform resharing.

• Amend vendor, agency and influencer agreements for SGI-use warranties, disclosure and indemnities.

• Make sure to check SSMI declaration requirements for each platform prior to upload to avoid rejection or delay.

• Expand the labelling and metadata requirements for platforms under the control of the brands and which host content created by users (reviews, contests, testimonials).

• Keep a centralised, dated compliance record of statements of use of AI, labels that have been applied and approvals that have been granted, that can be used in future ASCI and/or regulatory proceedings.

• Monitor finalisation of ASCI guidelines and judicial challenges to the IT Rules Amendment, 2026 and review internal policy.

Conclusion

The 2026 Indian rules governing the creation and dissemination of synthetic content represent a paradigm shift from the notice-and-takedown based approach to platform liability, to a more proactive approach of declaration-and-verification, which is supplemented by an advertising-specific overlay in the form of ASCI. In practice, for brands, AI governance can no longer be limited to creative or tech teams and needs to be integrated into labelling, provenance tracking, platform statements, and legal/compliance teams as well, giving ownership to these groups as well. Brands would be wise to take the most conservative compliant posture possible while the ASCI guidelines are being finalised and this constitutional validity of the IT Rules amendments are tested in court, recording their use of AI at the time to ensure compliance in the future.

Author: Prince Lucky Jain in case of any queries please contact/write back to us via email to content@khuranaandkhurana.com or at  Khurana & Khurana, Advocates and IP Attorney