On October 5, 2026, Civil Judge (Junior Division) Kuldeep Singh in Chandigarh ordered Meta, ex parte, to disable three Facebook pages (Batale Di Rajneeti, Kachi Laan and Thanedar Uncle) until October 22. The pages had posted AI-generated videos of Congress politician Amandeep Jaintipur of Batala, reportedly portraying him as "anti-social" and a "drug dealer". According to MediaNama's report, the videos had drawn about 3 lakh views. The court saw a prima facie case of defamation and observed that uncontrolled AI manipulation "may adversely affect the health of our democracy" and amounts to "a process of manufacturing public consent".
The case for the court
The strongest argument for this order is practical. A fabricated video that shows a candidate as a drug dealer does its damage in days, and a full defamation trial takes years. Punjab's elections are expected in early 2027, so a remedy that arrives after the vote is no remedy. The pages were also anonymous. The plaintiff could not identify the operators, and the platform was the only party able to act. An interim injunction against a platform is the ordinary tool for that situation. Judges who see a plausible, serious harm and a closing window will tend to act.
This is also not a one-off. MediaNama counts it as the third order from the Chandigarh Tricity area against AI-manipulated political content in about a year. In October 2025 a Mohali court ordered removal of 166 URLs carrying a fake video of Chief Minister Bhagwant Mann. On August 12, 2026, a Chandigarh civil court directed Meta to remove an AI-generated video of MLA Sukhpal Singh Khaira within 72 hours, finding it apparently defamatory and likely to cause irreparable reputational harm.
Why the escalation matters
The orders are getting broader. The Khaira order targeted a specific video. The Mann order covered identified URLs. The Jaintipur order suspends whole pages, including everything else they publish, until October 22. It also restrains the defendants from posting any further AI-generated content involving the plaintiff, which is an injunction covering future speech that the court has not yet seen.
That is where proportionality starts to matter. Indian free-speech law has long distrusted prior restraint and vague standards. In Shreya Singhal v. Union of India (2015), the Supreme Court struck down Section 66A of the IT Act for violating Article 19(1)(a). An ex parte order lasting 16 days is not Section 66A, and it is temporary and judicial. But a page-level suspension decided without hearing the page operators removes lawful commentary, satire and ordinary political criticism along with the defamatory videos. A court can limit this harm by ordering removal of identified videos and giving the operators a prompt chance to be heard.
The rules already on the books
The courts are not working in a vacuum. The government notified the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, 2026 on February 10, 2026, in force from February 20. Law-firm summaries, such as Hogan Lovells', describe the package. It defines "synthetically generated information" and requires prominent labels and embedded provenance metadata. Takedown of content flagged by a court or the government is due within three hours, down from 36.
The Khaira order shows the gap between those rules and practice. MediaNama noted that the video there reportedly carried none of the required AI disclaimers. The labelling regime put the compliance burden on platforms, but the harm arrived anyway, and a plaintiff still had to go to a civil court. The Election Commission's January 17, 2025 advisory asked political parties to label AI content. That binds parties, but it does little about anonymous "meme pages" that are not formally affiliated with anyone.
The Internet Freedom Foundation argues the 2026 rules define synthetic content so broadly that filters and translations fall in with harmful deepfakes, and that they rely on detection tools that perform poorly in multilingual settings. We think those concerns are fair. A blunt labelling mandate that cannot tell a harmful deepfake from a translated clip will leave the real fakes unlabelled and burden ordinary users.
A narrower path
The policy lesson is not that courts should do nothing, or that they should do everything. Three changes would help.
- Target the content, not the speaker. Orders should name specific URLs or videos, as the Mann and Khaira orders did, and leave the rest of a page running. Page suspensions should be a last resort for pages that exist only to spread the impugned content.
- Hear the other side fast. Ex parte relief is defensible when urgency is real. It is only legitimate if a short return date follows and the operators can come forward to contest it. A 16-day window is reasonable. Open-ended injunctions would not be.
- Fix detection before widening mandates. If the three-hour clock applies to court orders, platforms need a clear way to match a fake video to every re-upload. Without it, courts will keep issuing URL-by-URL orders that the next upload sidesteps.
India's regulators and judges are right to take election deepfakes seriously. Voters need to know what they are looking at, and candidates deserve a remedy that works before polling day. But a framework built on ever-broader ex parte orders, with labelling rules that miss the worst content, protects neither speech nor democracy well. Narrow, fast and reviewable orders do both.