---
title: "India's 3-hour takedown rule: AI audio and video must now carry a label"
summary: "What G.S.R. 120(E) asks of AI tools, large platforms and the companies that publish through them, read from MeitY's gazette text and FAQs, in force 20 February 2026."
url: https://www.thetechbag.com/blog/india-deepfake-rule-label-ai-media-three-hours
format: Analysis
desk: "AI in the enterprise"
published: 2026-02-20
facts_checked: 2026-02-20
author: TechBag (The TechBag Journal)
tags: ["IT Rules","deepfakes","synthetic media","AI labelling","content provenance","MeitY","G.S.R. 120(E)","EU AI Act","Article 50","China AI labelling","TAKE IT DOWN Act"]
cover: https://www.thetechbag.com/blog/covers/india-deepfake-rule-label-ai-media-three-hours.png
---

# India's 3-hour takedown rule: AI audio and video must now carry a label

*What G.S.R. 120(E) asks of AI tools, large platforms and the companies that publish through them, read from MeitY's gazette text and FAQs, in force 20 February 2026.*

Analysis · AI in the enterprise · By TechBag · Published 2026-02-20 · Facts checked 2026-02-20 · Canonical: https://www.thetechbag.com/blog/india-deepfake-rule-label-ai-media-three-hours

The short version

- From today, 20 February, an Indian platform that receives a court order or a reasoned government intimation must remove the content within 3 hours, not 36. A person’s complaint about intimate or morphed images of them gets 2 hours, not 24.[1][2]
- Realistic AI-made or AI-altered audio, images and video must carry a prominent label and, where technically feasible, permanent metadata with a unique identifier. Text is outside the definition, and so are routine edits, slide decks, subtitles and translation.[1][2]
- The duties fall on intermediaries: the AI tools that make such media, and platforms with 50 lakh or more users, which must ask uploaders to declare it and then check. A company publishing AI video is bound through them; one that lets its own users make or post it is an intermediary itself. Europe’s Article 50, from 2 August, binds the publisher directly.[1][6][8]

India now has a binding rule for AI-generated media. G.S.R. 120(E), notified by the Ministry of Electronics and Information Technology (MeitY) on 10 February, amends the IT Rules, 2021 and takes effect today, ten days later.[1] The draft appeared on 22 October 2025 and comments closed on 13 November.[4][5] The question for a business is narrower than the coverage of deepfakes suggests: if your marketing team makes an AI video, your helpline speaks in a synthetic voice, or your product lets customers generate either, what do you now have to do, and how fast?

## Three hours, not 36: four response clocks got shorter

The largest change is in Rule 3(1)(d). When an intermediary gets “actual knowledge” of unlawful content, which under the rule arises only from a court order or a reasoned written intimation from an authorised officer, it must now remove or disable access “within three hours”. The rule had said thirty-six hours.[1][3] MeitY’s own example: an intimation received at 11:00 AM must be complied with by 2:00 PM.[2]

3 hours

to take down content named in a court order or an authorised officer's intimation, from 20 February 2026; it was 36

Rule 3(1)(d), IT Rules, 2021, as amended by G.S.R. 120(E).[1]

Three other limits move with it. Under Rule 3(2)(b), a complaint from a person about content that exposes them, shows them in nudity or a sexual act, or impersonates them “including artificially morphed images” must now be acted on in 2 hours instead of 24. Removal requests under the proviso to Rule 3(2)(a)(i) fall from 72 hours to 36, and other grievances must be resolved in 7 days instead of 15.[1][3]

**Figure: The order-based takedown clock is now a twelfth of what it was; the morphed-image clock too**

*Hours an intermediary has to act, before and after G.S.R. 120(E), India, in force 20 February 2026*

|  | Before 20 Feb 2026 | From 20 Feb 2026 | Change |
| --- | --- | --- | --- |
| Court or officer order | 36 | 3 | -92% |
| Intimate or morphed image | 24 | 2 | -92% |
| Rule 3(1)(b) removal ask | 72 | 36 | -50% |

Source: G.S.R. 120(E), Gazette of India, 10 Feb 2026; MeitY FAQs, 10 Feb 2026 · as of 20 February 2026 · the fourth change, grievance resolution, moves from 15 days to 7 and is left off the scale

TechBag countTechBag count: four time limits change, and all four get shorter

We listed every substitution of a time period in the notification’s amendments to Rule 3: four in all. The cuts run from half (72 hours to 36) to eleven-twelfths (36 to 3, and 24 to 2). None is lengthened.

The 3-hour clock is not limited to deepfakes; it covers any unlawful content named in an order. The safeguards on who may issue one came three months earlier: G.S.R. 775(E), in force since 15 November 2025, requires an officer not below Joint Secretary (Deputy Inspector General for the police), a stated legal basis and the exact URL, and a monthly review by a Secretary.[7] The new rule lets a police administration authorise more than one such officer.[1] The case against is serious. The Internet Freedom Foundation, a day after notification, called the shortened deadline “an impossibly short timeline, which eliminates any meaningful human review”.[10]

## What must carry a label: realistic audio, images and video, not text

The new term is “synthetically generated information”, and the definition is narrower than the word “deepfake” suggests.

ReportedThe definition, Rule 2(1)(wa), verbatim

“audio, visual or audio-visual information which is artificially or algorithmically created, generated, modified or altered using a computer resource, in a manner that such information appears to be real, authentic or true and depicts or portrays any individual or event in a manner that is, or is likely to be perceived as indistinguishable from a natural person or real-world event”[1]

Three things are carved out: routine or good-faith editing that does not change the meaning; documents, presentations, training material and research outputs that do not create a false document; and tools used only for accessibility, translation, captions or search.[1] MeitY’s FAQs say plainly that “pure text / written outputs, by themselves, are not SGI”, and give examples on both sides: an AI-designed slide deck or subtitles are not covered; a realistic AI presenter, or a synthetic voice that reproduces a real person’s, is.[2]

Covered content then splits in two. Four kinds must be blocked outright: sexual abuse and non-consensual intimate imagery, false documents or records, material on explosives or arms, and content that falsely depicts a real person or event in a way likely to deceive. Everything else must be labelled so it is “easily noticeable”, with audio carrying “a prominently prefixed audio disclosure”, and embedded with permanent metadata, including a unique identifier, “to the extent technically feasible”. The intermediary may not let anyone remove either.[1] The FAQs spell out the last part: no “remove watermark” or “export without metadata” option.[2]

**Figure: Between draft and rule, the scope narrowed, the fixed label size went, and the takedown clocks arrived**

*MeitY's draft amendment of 22 October 2025 against the notified rule of 10 February 2026, key provisions*

Draft, 22 October 2025

- Any “information” that reasonably appears authentic or true, text not excluded
- Label covering at least 10% of the visual display, or the first 10% of audio
- No carve-outs for routine edits, documents or translation
- Takedown and grievance clocks left as they were

Notified rule, in force 20 February 2026

- Only audio, images and video that look real and depict a person or event
- A “prominent” label, a spoken prefix for audio, no fixed size
- Three explicit exclusions; text-only output not covered
- 3 hours, 2 hours, 36 hours and 7 days, all shorter

Source: MeitY draft amendment and explanatory note, 22 Oct 2025; G.S.R. 120(E), 10 Feb 2026 · as of 20 February 2026

The draft’s 10% rule, which would have put a label over a tenth of every frame, is gone.[4] So is any doubt about text: an AI-written blog post or chatbot reply is not synthetic information under these rules, though it can still be unlawful information under the older ones.[2]

## Who it binds: the tool, the platform, and your company through both

The rules regulate intermediaries, and two kinds carry the new duties. Rule 3(3) applies to any intermediary offering a computer resource that can create, alter, publish or share synthetic content; MeitY lists AI image and video generators, voice-cloning tools and platforms that help publish such content.[1][2] Rule 4(1A) adds duties for significant social media intermediaries, those with 50 lakh or more registered users in India: before anything is published, ask the user to declare whether it is synthetic, “deploy appropriate technical measures” to check the declaration, and label it if confirmed.[1][6] A platform that knowingly lets unlabelled synthetic content through is deemed to have failed due diligence.

Two duties reach every intermediary, synthetic content or not. Users must be told of the platform’s rules and the penalties at least once every three months, where the FAQs say the earlier requirement was once a year.[1][2] And the price of non-compliance is unchanged: under Rule 7, an intermediary that fails to observe the rules loses the Section 79(1) protection from liability for its users’ content.[3]

**Figure: A company's AI video meets the rule three times: in the tool, at upload, and when someone complains**

*The path of one AI-generated marketing video under the amended IT Rules, India, from 20 February 2026*

1. 01Made in an AI toolThe tool labels the video and embeds metadata with an identifier for the tool used, and may not offer a way to strip either.
2. 02Uploaded to a large platformThe uploader declares it synthetic; a platform with 50 lakh+ users checks the claim and shows a label before it goes live.
3. 03Someone complains2 hours if it impersonates or morphs the person complaining; 36 hours for other Rule 3(1)(b) removal requests; 7 days otherwise.
4. 04An order arrivesA court or an authorised officer names the content: 3 hours to remove or disable it.

Source: G.S.R. 120(E), 10 Feb 2026 (Rules 3(1)(d), 3(2), 3(3), 4(1A)); MeitY FAQs, 10 Feb 2026 · as of 20 February 2026

Our view

Our reading, not legal advice: the rules do not name advertisers or brands. A company that publishes its own AI video on its own website is the publisher of that content, not an intermediary for it, so Rule 3(3) does not bind it directly. It is bound in practice by the tool it used and the platform it posts on. The exception matters for software firms: if your product lets customers generate AI voices or avatars, or host and share media, you are the intermediary, and every duty above is yours.

## Europe binds the publisher, China labels text, India binds the platform

The European Union’s version applies later and reaches further. Article 50 of the AI Act, Regulation (EU) 2024/1689, applies from 2 August 2026. Providers of systems that generate synthetic audio, image, video or text must mark the output in a machine-readable format, detectable as artificially generated. Deployers, meaning the businesses that use such a system, must disclose a deep fake they publish; evidently creative or satirical work gets a lighter disclosure, and AI text published to inform the public on matters of public interest must be disclosed unless a person has reviewed it and holds editorial responsibility. Fines reach €15 million or 3% of worldwide annual turnover, whichever is higher.[8]

Two pieces of the EU picture are still moving. The Commission published the first draft of a code of practice on marking and labelling on 17 December 2025.[9] Its Digital Omnibus proposal of 19 November 2025 would give generative systems already on the market before 2 August 2026 until 2 February 2027 to add marking; that is a proposal before the Parliament and Council, not law.[11][12]

China went first. Its Measures for Labelling AI-Generated Synthetic Content (人工智能生成合成内容标识办法), issued by the Cyberspace Administration with three other agencies, have applied since 1 September 2025: a visible label on text, audio, images, video and virtual scenes, plus a hidden label in the file’s metadata naming the provider and a content ID.[13] The United States has taken the takedown route instead: the TAKE IT DOWN Act of 19 May 2025 gives platforms 48 hours to remove intimate images, including software-made “digital forgeries”, after a valid request, with the process due within a year of enactment.[14] India’s clock for the same harm, on the complaint of the person shown, is 2 hours: a twenty-fourth of that.[1]

**Figure: India's rule arrives nearly six months after China's and five and a half before Europe's**

*Dated milestones in AI-media labelling and takedown rules, India, the EU, China and the US, May 2025 to August 2026*

| Date | Event | Detail |
| --- | --- | --- |
| 19 May 2025 | US: TAKE IT DOWN Act | 48 hours to remove intimate images, including digital forgeries |
| 1 Sept 2025 | China: labels in force | visible and metadata labels, text included |
| 22 Oct 2025 | India: draft published | label to cover 10% of the frame |
| 19 Nov 2025 | EU: omnibus proposal | marking grace to 2 Feb 2027 for existing systems, proposed |
| 17 Dec 2025 | EU: first draft code | code of practice on marking and labelling |
| 10 Feb 2026 | India: G.S.R. 120(E) | notified in the Gazette |
| 20 Feb 2026 | India: rule in force | labels, declarations, 3-hour takedown |
| 2 Aug 2026 | EU: Article 50 applies | marking by providers, disclosure by deployers |

Source: Gazette of India; MeitY; Regulation (EU) 2024/1689; European Commission; CAC; US Public Law 119-12 · as of 20 February 2026 · 2 August 2026 is the application date fixed in the AI Act as published; the omnibus entry is a proposal

Our view

Our view: India made the platform the checkpoint; Europe makes the business that publishes answerable as well. An Indian company with European customers should plan for both: platform labels at home from today, and its own disclosure on deep fakes it publishes in the EU from 2 August.

## What a business should change now

Most of the work is an inventory and a few contract lines. The order below starts with what every company that publishes AI media can do this week and ends with the duties that apply only if you run a platform.

**Figure: Six changes cover a company that publishes AI media and one that lets users make it**

*Actions for Indian businesses under the amended IT Rules, TechBag's sequence, February 2026*

Step 1 of 6

1. List where you make AI media

Ads, training videos, product demos, AI presenters, the helpline’s synthetic voice. Text and slide decks can stay off the list; realistic audio, images and video go on it.

Source: TechBag analysis of G.S.R. 120(E) and MeitY FAQs, 10 Feb 2026 · as of 20 February 2026

If you are buying this in India

There is no fee, licence or registration in G.S.R. 120(E); the cost is operational, and it lands on whoever is the intermediary.[1] When you buy an AI image, video or voice tool for use in India, ask four questions before you sign. Does it label output and embed metadata with a unique identifier, as Rule 3(3) requires? Does an export keep both? Does it warn users at onboarding, as Rule 3(1)(ca) now requires of tools that make synthetic content? And who is its grievance officer in India, and how does it meet the 2-hour and 3-hour clocks?[1][3]

In the contract, ask for an indemnity if the tool’s labelling fails, and for notice before any feature that removes labels is switched on. If your own product hosts user media, treat 20 February as a go-live date already passed: the clocks above apply from today, and MeitY’s FAQs, which describe themselves as an evolving document, are not a substitute for the rule text.[2]

How we did this

Figures as of 20 February 2026

We read the English text of G.S.R. 120(E) in the Gazette of India, Extraordinary, No. 114 of 10 February 2026, MeitY’s FAQs and consolidated IT Rules of the same date, the October 2025 draft and explanatory note, G.S.R. 775(E) and S.O. 942(E), all from MeitY’s site. Time limits are quoted from the rule text; the FAQs are cited only for MeitY’s examples, and say themselves that they are not a legal document. The count of changed time limits is ours, taken from the substitutions in paragraph 3 of the notification. The EU material is from Regulation (EU) 2024/1689 and the Commission’s pages; the omnibus proposal’s Article 50 date is from CMS’s summary, because the proposal on EUR-Lex could not be opened. China’s measures were read in Chinese on the Cyberspace Administration’s site. Who is bound in Section 3 is our reading, not legal advice. MeitY publishes no list of significant social media intermediaries, so we give no count.

Spotted an error? Write to us; corrections are published at the foot of the piece.

Sources

1. [MeitY, G.S.R. 120(E): Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, 2026, Gazette of India Extraordinary No. 114 (also egazette.gov.in/WriteReadData/2026/269993.pdf)](https://www.meity.gov.in/static/uploads/2026/02/f55fe52418b03f58b0669f6a8bc03b6d.pdf) — 10 February 2026 In force 20 February 2026.
2. [MeitY, Frequently Asked Questions on the IT (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, 2026](https://www.meity.gov.in/static/uploads/2025/10/065b6deb585441b5ccdf8be42502a49c.pdf) — 10 February 2026
3. [MeitY, IT (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, consolidated text updated 10.02.2026 (Rules 3(1)(d), 3(2), 7)](https://www.meity.gov.in/static/uploads/2026/02/550681ab908f8afb135b0ad42816a1c9.pdf) — 10 February 2026
4. [MeitY, Draft amendments to IT Rules, 2021 relating to synthetically generated information, and explanatory note](https://www.meity.gov.in/static/uploads/2025/10/9de47fb06522b9e40a61e4731bc7de51.pdf) — 22 October 2025 Explanatory note: meity.gov.in/static/uploads/2025/10/8e40cdd134cd92dd783a37556428c370.pdf
5. [MeitY, Notice extending the last date for comments on the draft to 13 November 2025](https://www.meity.gov.in/static/uploads/2025/10/99452ba2c2ce83f525c0fd6929a879d7.pdf) — 6 November 2025
6. [MeitY, S.O. 942(E): threshold of fifty lakh registered users in India for a significant social media intermediary](https://www.meity.gov.in/static/uploads/2025/10/e671ed889878ed9540903e8b23fce65f.pdf) — 25 February 2021
7. [MeitY, G.S.R. 775(E): IT Amendment Rules, 2025, substituting Rule 3(1)(d) (officer rank, reasoned intimation, monthly review)](https://www.meity.gov.in/static/uploads/2025/10/90dedea70a3fdfe6d58efb55b95b4109.pdf) — 22 October 2025 In force 15 November 2025.
8. [Regulation (EU) 2024/1689 (Artificial Intelligence Act), Article 50 (transparency), Article 99(4) (fines) and Article 113 (application)](https://eur-lex.europa.eu/eli/reg/2024/1689/oj) — 13 June 2024
9. [European Commission, Code of Practice on transparency of AI-generated content: first draft](https://digital-strategy.ec.europa.eu/en/policies/code-practice-ai-generated-content) — 17 December 2025
10. [Internet Freedom Foundation, IT Intermediary Amendment Rules, 2026 contradict their purpose](https://internetfreedom.in/it-intermediary-amendment-rules-2026-contradict-their-purpose/) — 11 February 2026
11. [European Commission, Digital Omnibus on AI Regulation proposal, COM(2025) 836](https://digital-strategy.ec.europa.eu/en/library/digital-omnibus-ai-regulation-proposal) — 19 November 2025
12. [CMS, Digital Omnibus on AI: adjusted timelines and transitional provisions explained](https://cms.law/en/gbr/publication/digital-omnibus-on-ai-adjusted-timelines-and-transitional-provisions-explained) — 17 December 2025
13. [Cyberspace Administration of China, MIIT, Ministry of Public Security, NRTA: 人工智能生成合成内容标识办法 (Measures for Labelling AI-Generated Synthetic Content), Articles 3 to 5 and 14](https://www.cac.gov.cn/2025-03/14/c_1743654684782215.htm) — 14 March 2025 In force 1 September 2025.
14. [United States, TAKE IT DOWN Act, Public Law 119-12, sections 2 and 3](https://www.govinfo.gov/content/pkg/PLAW-119publ12/html/PLAW-119publ12.htm) — 19 May 2025

Related on TechBag: [Canva AI](https://www.thetechbag.com/canva/canva-ai) and [Adobe Creative Cloud](https://www.thetechbag.com/adobe/adobe-creative-cloud) are two of the image and video generators whose output now needs a label in India; the [email security guide](https://www.thetechbag.com/browse/security/email-security) covers the impersonation filters that sit in front of a deepfaked request for money.

**Disclosure.** TechBag is a software reseller: it earns a margin when a reader buys through it, including from vendors named here. The analysis is TechBag’s own; no vendor paid for or saw this piece before publication. Facts are labelled reported, TechBag count or our view.

Go deeper on TechBag

- [Canva AI](https://www.thetechbag.com/canva/canva-ai) — an image and video generator whose output now needs a label in India
- [Adobe Creative Cloud](https://www.thetechbag.com/adobe/adobe-creative-cloud) — Firefly generation across the apps, and the exports that must keep their metadata
- [Email & collaboration security guide](https://www.thetechbag.com/browse/security/email-security) — the impersonation filters in front of a deepfaked request for money