India's 2026 AI labelling rules: what changes for you

In February 2026, India got its first rules written specifically for AI-generated media. The Ministry of Electronics and IT (MeitY) amended the IT Rules, 2021, to require labels on AI-made content and much faster removal of unlawful posts, including deepfakes. Here's what they say, based on published legal summaries, and what they mean for ordinary users.

This is a plain-language summary, not legal advice. For the exact wording, read the notification itself (G.S.R. 120(E)).

The basics

• Notified: 10 February 2026, as an amendment to the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021.
• In force: from 20 February 2026, according to Hogan Lovells and other law-firm summaries.
• Covers: "synthetically generated" content, meaning audio, images or video made or altered by computer so that it appears authentic.

What platforms now have to do

• Label AI-made content prominently, and attach provenance metadata that identifies it as synthetic.
• Ask users to declare AI content. Large platforms, those with more than 50 lakh users in India, must ask uploaders whether their content is AI-generated and use technical checks before it's published, as Business Today reported.
• Take down unlawful content much faster: within 3 hours of a court or government order (down from 36 hours), and within 2 hours of a complaint about non-consensual intimate imagery, including morphed and deepfake nude images.

What it means for you

Expect more "AI-generated" labels on Instagram, YouTube, Facebook and X. They help, but don't read too much into a missing label: labels depend on uploaders being honest and on platforms detecting AI correctly, and provenance metadata is lost when someone screen-records or re-encodes a video. An unlabelled video isn't proof that it's real, which is why checking the video itself still matters.

If someone posts a deepfake of you, especially an intimate one, report it to the platform straight away using its reporting tools or its grievance officer (every major platform must list one). The 2-hour rule for intimate imagery is on your side. Also report it at cybercrime.gov.in or 1930, and keep the links and screenshots. Our guide on what to do if someone made a deepfake of you has the full steps.

If you create content with AI, disclose it when you upload. Platforms are now required to ask, and an honest label protects you if the clip is later reported.

What the rules can't do

Rules make platforms label and remove content faster; they don't stop a fake from spreading in WhatsApp groups before anyone reports it. Forwarded videos lose their labels and metadata along the way. The habit that still protects you most is the oldest one: check before you share.

Try it on the next forward that feels off: check a video free, or see how good your own eye is with the Real or AI quiz.