The Delhi High Court has restrained AI deepfakes and fabricated speeches of Kangana Ranaut and set 72-hour takedown and blocking directions for Meta, MeitY and the Department of Telecommunications

Key takeaways

• The relief is interim and ex-parte. The Court found a prima facie strong case and said the balance of convenience favours the plaintiff.
• Meta is directed to remove and permanently take down the listed content within 72 hours of the plaintiff sending a copy of the order, and to file a compliance affidavit within two weeks.
• MeitY and the Department of Telecommunication are directed to issue blocking directions to ISPs and TSPs within the same 72 hours.
• The plaintiff may notify further URLs, accounts or channels, and the defendants are to remove them without fresh proceedings.

Is this a final ruling, and who does it bind?

No. The order was passed on the plaintiff's applications under Order XXXIX Rules 1 and 2 CPC and for a dynamic injunction, and the Court granted relief "till the next date of hearing". The suit was registered, summons were issued, and the matter is listed before the Court on 3 February 2027, with service and pleadings before the Joint Registrar on 4 December 2026. The Court recorded the plaintiff's case as stated in the plaint, including that she is an actor, producer and sitting Member of Parliament for Mandi, and its own view is expressed as prima facie. The order was passed ex parte and records no written statement from the defendants (it sets 30 days for defendants 10 and 11 and thirty days from receipt of summons for the others), so it does not reflect their position. Two defendants, nos. 6 and 7, were deleted from the array of parties after the plaintiff's counsel said they had taken down the offending material.

What did the Court restrain?

Paragraph 64(I) restrains the defendants, including John Does, and those acting for them, from three things: exploiting the plaintiff's name, image, likeness, voice, speeches, signature and persona without authorization; "creating, publishing, disseminating or communicating any AI-generated content, deepfakes, synthetic media, manipulated videos, manipulated audio recordings, fabricated speeches, false endorsements or impersonation content" depicting or purporting to depict her; and operating any account, page, profile, channel, website or digital asset falsely representing association, endorsement or authorization by her.

On the reasoning, the Court said the plaintiff's stature gives her proprietary rights over her personality attributes. It said protection extends to AI-generated or morphed sexual and explicit deepfakes, and to AI-generated content that falsely shows her making statements or attributing false quotes to her, which it called inherently deceptive. It treated Document K as ex-facie obscene, morphed or AI-generated content and said a specific image at page 67 of Document G, which the plaintiff said a fact-check report certified as fake, was liable to be taken down. The plaintiff did not press for removal of a satirical meme, which Annexure-A leaves out.

What must Meta, MeitY and the Department of Telecommunication do, and when?

Meta, described in paragraph 64(II) as defendant no. 2/Meta, is directed to "remove, disable access to, and permanently take down" the content in Annexure-A and Document K, including videos, reels, posts, accounts or channels that unlawfully use or imitate the plaintiff's persona, within 72 hours of being communicated a copy of the order by the plaintiff. A compliance affidavit is to be filed within two weeks.

MeitY (defendant no. 10) and the Department of Telecommunication (defendant no. 11) are directed to issue necessary notifications or directions to all ISPs and TSPs for blocking or removing access to the infringer websites in Annexure-A and Document K, within 72 hours of the same communication, with their own compliance affidavit in two weeks. Both accepted notice through central government counsel and have 30 days for a written statement. Annexure-A lists 18 items, mostly Instagram posts, reels and searches, and several entries are marked removed or renamed. The order is not consistent on numbering: Annexure-A labels Meta Platforms, Inc as defendant no. 1 and uses defendant no. 2 for an account holder (Golu Yadav), while paragraph 64(II) calls Meta defendant no. 2. We follow paragraph 64(II) for the directions and flag the mismatch.

The order also leaves room to adjust. Defendants may ask how a listed URL is covered. A website or account that is not primarily infringing, and is blocked under the order, may approach the Court with an undertaking not to infringe her personality rights, and the Court would then consider modifying the injunction. The plaintiff may notify additional infringing URLs, accounts, channels or content found after filing, and the defendants "shall thereupon remove or disable access to the same without requiring the plaintiff to institute fresh proceedings".

What does it mean for platforms and media counsel, and what should they do now?

The order cites IT Rules, 2021 twice. Paragraph 56 says Rule 3(1)(b)(ii) makes it obligatory on an intermediary not to host or transmit obscene or pornographic information, and paragraph 61 says the objection to the false quotes would fall within Rule 3(1)(b)(v). It does not decide a general duty to detect AI content, and it does not discuss safe-harbour loss. For a platform, the practical point is operational: a named-person order with a 72-hour clock and a standing right to add URLs.

• Record the date and time a copy of the order reaches your legal or grievance channel, since the clock runs from the plaintiff's communication.
• Map each Annexure-A and Document K item to a takedown action and keep evidence of removal for the compliance affidavit.
• Set a route for additional URLs the plaintiff notifies later, and a route for a wrongly blocked account to seek modification.
• If you advise a public figure, note that the Court relied on a sworn plaint, a fact-check report and a tabulated link list, so the supporting record mattered.

What we did not verify

What we opened: the full 19-page order of 8 October 2026, including Annexure-A, as published by LiveLaw from the Delhi High Court server (download stamp 12 October 2026), on 24 October 2026.

What we did not open: the plaint, Document K, Document G, the fact-check report, the Delhi High Court Order Portal copy, any compliance affidavit, and any later order in the suit. We did not check whether the listed content has been removed. The plaintiff's account of the facts is the plaintiff's contention and we have not tested it. We rely on the order's own date of 8 October 2026; press reports are dated later.

What we refuse to claim: that the order sets a rule for all platforms, that any platform lost safe-harbour protection, that any content was in fact removed, or what the Court will decide at the final stage.

Related briefings

• AI Regulation News hub
• Asia-Pacific AI regulation tracker
• Attorneys and AI regulation
• Delhi High Court order on AI deepfakes of Ravi Kishan and personality rights
• India IT Rules amendments on synthetic media labelling
• MeitY AI governance guidelines and actor liability
• Bar Council of India circular on social media, AI and deepfakes for advocates