AI-Generated Ads: New India Rules 2026 – Every Digital Marketer Must Know
By Vaibhav Prashar, Digital Marketing Trainer in Dharamshala and Kangra, HP
If you use AI to create ads, there are now specific rules governing this. This guide explains everything in detail.
Written by Vaibhav Prashar
Digital Marketing and AI Trainer, Dharamshala. MESC and RASCI Certified. 9+ Years. 7000+ Students.
I am Vaibhav Prashar, a certified digital marketing trainer based in Dharamshala and Kangra, Himachal Pradesh. Over the past 9 years, I have trained more than 7,000 students in digital marketing, AI tools, and communication skills.
When I teach my students how to run social media ads and use AI tools, I feel it is essential for me to stay updated on new regulations myself. That is why I researched ASCI new AI-advertising disclosure rules and put together this complete guide.
1. Why These Rules Came About
Companies use AI in several ways to create ads – creating a virtual model instead of a real one, generating an AI version of a celebrity voice, enhancing product photos until they look nothing like the real product, or generating fake customer reviews with AI.
When you see an ad, you naturally believe what you are seeing is real. If it is AI-generated and you cannot tell, that is a form of deception – you buy a product based on a false impression.
A Detailed Example
Imagine launching a face cream. Approach 1 (Honest): a real person uses the cream for 4 weeks, genuine before-after photos. Approach 2 (Misleading): an AI-generated after photo showing results the cream could never achieve.
Approach 2 is exactly what these new rules try to prevent.
MeitY (Ministry of Electronics and IT) updated its IT Rules in February 2026, defining Synthetically Generated Information (SGI). ASCI (Advertising Standards Council of India) then published its own guidelines on 8 May 2026.
2. The 3-Tier Risk Framework
Completely Prohibited
Label Required
No Label Needed
3. High Risk – Never Do This
1. Fabricated Testimonials
Example: Generating 20 fake customer reviews with AI when no such customers exist.
2. AI-Generated Doctors
Example: An AI-generated doctor character recommending your supplement.
3. Exaggerated Claims
Example: A weight-loss ad showing an AI-generated 15kg-in-30-days result that is never achievable.
4. Fabricated Locations
Example: An AI-generated factory photo claiming your product is made there.
5. Deepfakes Without Consent
Example: Using an AI-generated celebrity voice or face without their permission.
4. Medium Risk – Label Required
Synthetic Influencers
Example: A fully AI-generated influencer character – allowed, but must disclose it is AI.
The Consent Myth
Even with 100 percent celebrity consent, disclosure is still required – this rule is about telling the consumer the truth, not about permission.
5. Low Risk – No Concern
Routine color correction, lighting adjustments, and AI-assisted copywriting need no disclosure.
8. Legal Consequences Explained
Section 79 IT Act
Platforms normally are not liable for user content (safe harbor). If a platform fails to label AI content properly, it can lose this protection and become directly liable.
CCPA and Consumer Protection Act
CCPA can impose fines and order corrective advertising for misleading AI-generated ads.
8.5 A Real Case – Shilpa Shetty Deepfake
The Bombay High Court ordered removal of AI-generated deepfake content involving actress Shilpa Shetty Kundra, based on personality rights – every person right to their own image, voice, and identity.
8.6 Real Enforcement Numbers
ASCI processed 1,409 influencer violation cases by November 2025 – 94 percent were disclosure failures. CCPA can fine individuals up to 10 lakh rupees and entities up to 50 lakh rupees, plus a 3-year endorsement ban.
9. Practical Checklist
- Categorize each AI ad by risk-tier first
- Completely avoid High-Risk tactics
- Add clear AI labels on Medium-risk ads
- Never skip disclosure even with consent
- Add verbal disclosure in first 10 seconds of video
- Review ASCI guidelines quarterly
10. FAQs
Is AI advertising banned now
No, only specific high-risk misleading tactics are banned.
Are these rules legally binding
Currently draft guidelines, advisory in tone, but MeitY IT Rules are already binding.