AI-Generated Content Labelling Rules in India
AI-generated content labelling means clearly telling viewers when an image, video, voice or text was created or changed by AI, especially when it looks real. In India this is shaped by the IT Rules on synthetic media, government advisories, platform policies from Meta and YouTube, and ASCI's advertising code. This page is general information, not legal advice.
- Why labels: To stop people being misled by realistic fakes, such as deepfake videos and cloned voices.
- Who must act: Platforms carry legal duties, and they pass them on to creators and advertisers through their policies.
- What counts most: Realistic people, voices and events that did not happen, not simple edits like colour correction.
- Consent: Using a real person's face or voice needs their permission, with or without a label.
- Ads: ASCI's rule against misleading ads applies however an ad is made.
This lesson follows one example: a saree brand from Surat planning its Diwali sale. It wants AI-generated model photos in five saree colours, a short video of an AI model walking through a festive set, and an AI voiceover in Gujarati and Hindi. It will run the ads on Instagram, Facebook and YouTube.
Key Characteristics of AI-Generated Content Rules in India
- Built on the IT Rules: The rules sit in the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, as amended.
- Advisories came first: MeitY issued advisories on deepfakes and on labelling AI content in late 2023 and 2024.
- Formal rules followed: The government then amended the IT Rules to cover synthetically generated information, including labels and user declarations on large platforms.
- Platform policies go further: Meta and YouTube have their own disclosure rules that apply to every post and ad.
- Advertising code applies: ASCI's code and consumer protection law forbid misleading ads, whatever tool made them.
How AI-Generated Content Labelling Works
- Creator declares: When uploading, the brand tells the platform the content is AI-generated or altered, using the platform's setting.
- Platform checks: Large platforms may also detect AI content through technical signals such as embedded metadata, and add labels themselves.
- Label shown: Viewers see a label, such as "AI info" on Meta or an altered or synthetic content note on YouTube.
- Metadata kept: Some rules and platforms expect a permanent identifier or metadata in the file that cannot be removed easily.
- Complaints handled: If a label is missing or content harms someone, platforms must act on complaints within set times.
Platform Labels: Meta and YouTube
- Meta (Instagram, Facebook): Meta asks users to disclose photorealistic video or realistic audio that was made or changed digitally, and may add an "AI info" label. Ads about social issues, elections or politics have extra disclosure rules.
- YouTube: Creators must disclose realistic altered or synthetic content in YouTube Studio during upload. YouTube shows a label in the description or, for sensitive topics, on the video player.
- Google Ads: Election ads with synthetic content that looks real need a clear disclosure.
ASCI and Influencer Rules
ASCI's influencer guidelines ask virtual influencers to disclose to consumers that they are not interacting with a real human. If the saree brand uses an AI model as its face on Instagram, the profile and posts should say so. AI creators and avatars are covered in AI influencers and virtual creators. Product images must also show the real product: an AI image that changes a saree's colour, weave or drape can mislead buyers.
Example: Labelling a Diwali Campaign Made with AI
- Model photos: AI-generated models wearing the brand's real sarees, photographed flat first. Posts are marked as AI with the platform setting, and captions add "Model image created with AI".
- Video: The walking video looks realistic, so it is disclosed on YouTube and Instagram and carries a small on-screen line: "AI-generated video".
- Voiceover: A synthetic voice from a licensed AI tool, not a copy of any real person. The brand keeps the licence.
- Product truth: Every colour and border in the AI images is checked against the physical saree.
- Records: The team saves prompts, tool names, source photos and final files for each ad.
Practical Checklist for AI-Generated Content
- Did AI create or change it? Note the tool and what it did.
- Does it look or sound real? If yes, label it on the platform and in the caption or on screen.
- Is a real person shown or voiced? Get written consent, and never imitate a celebrity.
- Is the product shown truly? Colour, size and results must match reality.
- Is personal data used? Customer photos used to train or prompt AI need consent under the DPDP Act for marketers.
- Are records kept? Save prompts, files and approvals.
Benefits of Labelling AI Content
- Trust: Buyers are less likely to feel tricked when they see a label up front.
- Fewer takedowns: Correct disclosure lowers the risk of ads being rejected or posts removed.
- Ready for audits: Records make it easy to answer a platform, ASCI or a customer.
Limitations and Open Questions
- Rules still changing: Definitions, label sizes and timelines may change as rules and platform policies are updated.
- Detection gaps: Platforms cannot detect every AI file, so honest disclosure still matters.
- Grey areas: Light edits, like background clean-up, may not need a label, but the line is not always clear.
How AI Changes Ad Content Labelling
What AI Automates Now
AI tools can now make realistic models, sets and voices in minutes, which is why labelling matters. Some tools add content credentials or watermarks automatically, and platforms use AI to detect synthetic media. A large language model can also draft captions and disclosure lines.
What Still Needs a Human
Judging whether content looks real enough to need a label, getting consent from real people, and checking that the product is shown truly all need a person. See responsible AI in marketing for a wider policy.
Risk to Watch
Removing watermarks or metadata to avoid a label may itself break rules. AI UGC-style testimonials that look like real customers can mislead even with a label; AI UGC videos need disclosure and consent from anyone whose likeness is used.
Do It with AI
Use this prompt to review an AI-made campaign before it goes live. It works in ChatGPT, Claude or Gemini.
You are a compliance reviewer for a brand advertising in India. Campaign: [product, festival or sale, platforms] Assets made or changed with AI: [list each image, video, voice or text and the tool used] Real people involved: [models, staff, customers, none] 1. For each asset, say whether it looks or sounds realistic enough to need a label. 2. Suggest short label wording for the caption and on screen, in English and [language]. 3. List any asset that needs a real person's written consent. 4. Flag any way an asset could mislead buyers about the product. 5. List what I should check in the current IT Rules, platform policies and ASCI guidance. Do not state section numbers or deadlines as fact.
- List every asset in the campaign and how AI was used.
- Run the prompt and add labels where it flags realistic content.
- Collect consent forms and check product accuracy.
- Turn on the platform's AI disclosure setting when posting, and save all records.
Check Before You Use It
- Facts: Confirm the current rules on MeitY's site and each platform's help centre, not in an AI summary.
- Brand fit: Keep labels short and plain so they sit well in the ad.
- Compliance: Get consent from any real person shown or voiced, disclose AI avatars and influencers, and take legal advice for high-risk campaigns.
Quick Quiz
Pick an answer to check yourself. Nothing is saved.
Question 1 / 3
1. The Surat saree brand's Diwali video shows an AI-generated model who looks like a real woman. What should it do?
Frequently Asked Questions
Do I have to label AI-generated images in ads in India?
India has moved to require labels on synthetically generated content, mainly through duties on online platforms under the IT Rules, and platforms pass these duties on to advertisers and creators. Check the current notified rules and each platform's policy, and label realistic AI content when in doubt.
What is synthetically generated information?
It is the term India's IT Rules amendments use for content created or changed by a computer so that it looks real or true, such as a deepfake video or a cloned voice. Check the notified definition for its exact scope and exceptions.
How do I add an AI label on Instagram or YouTube?
Meta lets you mark content as made with AI when posting, and applies an AI info label. YouTube asks creators to disclose realistic altered or synthetic content in YouTube Studio when uploading. Menu names change, so check the platform's help pages.
Can a brand use an AI version of a celebrity in an ad?
Not without the person's written permission. Using someone's face, voice or name without consent can breach their personality rights, mislead consumers and break platform policies. Indian courts have granted orders protecting well-known people from such misuse.
Does ASCI have rules on AI in advertising?
ASCI's code applies to all ads, however they are made, so AI ads must not mislead. Its influencer guidelines also ask virtual influencers to disclose that they are not real people. Check ASCI's website for any newer AI guidance.
Related Articles
- Responsible AI in MarketingResponsible AI in marketing means disclosure, consent, fairness and privacy. Learn the principles and checks through a Kolkata kirana store example.
- AI UGC VideosAI UGC videos explained: what they are, when they are honest, and the rules on disclosure, consent and fake testimonials, with an Ahmedabad snack brand.
- AI Influencers and Virtual CreatorsAI influencers explained: how virtual creators are made, the disclosure and consent they need in India, and when they help a brand, with a fashion example.
- AI Image Generation for MarketingAI image generation for marketing: how it works, where it helps and the rules on rights, likeness and disclosure, with a Varanasi saree brand and a prompt.
- AI Video Generation for AdsAI video generation for ads explained: tools, a script to export workflow, platform policies and limits, with a Moradabad brass brand's Diwali sale ad.
- DPDP Act for MarketersDPDP Act for marketers explained simply: consent, notice, withdrawal, user rights and children's data, so your leads and campaigns stay on the right side.