When a Video Lies About You: Inside India's New Deepfake Takedown Rules



             
AI-generated illustration accompanying the article.

Source: Created by the author using OpenAI's ChatGPT image generation tool (2026).


Introduction: 

A student got a notification like any other, someone had tagged her in a video. When she opened it, she saw herself, saying things she never said, in her own voice, cloned to sound exactly like her. It was obviously AI-generated. That didn't make it feel any less like a violation. Nobody had asked her permission, and by the time she found the video, it had already been watched, shared, and forwarded well beyond her control. This kind of story isn't unusual anymore. Across social media, young creators are quietly changing how much they post, and how, worried that their face or voice could end up in a video they never made. 

The government has taken notice: the government's response is a set of amendments to the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, made under the Information Technology Act, 2000, updated as of February 10, 2026.

What the Rules Actually Say ?

At their core, the updated rules spell out how much diligence platforms (the intermediaries hosting our posts, photos and videos) now owe their users, and how complaints are supposed to be handled. They also lay out a code of ethics and a set of procedural safeguards specifically for digital media. The underlying goal is straightforward: block unlawful, synthetically generated content wherever it's reasonable to do so. In practice, the rules draw a sharp
line based on content type. For specified categories such as non-consensual intimate imagery, nudity, sexual content and certain impersonation or morphed content, intermediaries must remove or disable access within two hours of
receiving a complaint. Everything else, including a video like the one described above, gets a slower track: the platform has to acknowledge the complaint within 24 hours and resolve it within seven days. If a user isn't satisfied with how a platform handles their complaint, they can escalate it to a newly established Grievance Appellate Committee, which is required to decide the appeal within a month. On top of that, intermediaries now have to build
in a reporting mechanism and prompt users to disclose when content they're uploading is AI-generated.

So, How Do You Actually Get Something Taken Down ?

If you find yourself in a situation like this, the process looks roughly like this: preserve the evidence before anything else: screenshot it, save the link, and note the time you found it. The complaint should first be made through the platform’s grievance-redressal mechanism or to its Grievance Officer. State clearly that you (or the person you’re representing) are the one shown or heard in the content and formally request that it be taken down or disabled. Keep every acknowledgment the platform sends you. If the complaint is rejected or remains unresolved, it can be escalated to the Grievance Appellate Committee (GAC) through its online portal.

Where the Rules Fall Short ?

The two-tier system is the first crack. However, a few gaps stand out: 

● Sexual or obscene deepfakes get fast-tracked, but a fabricated video that defames someone, puts words in their mouth, or damages their reputation doesn't move any faster, even though the harm to a person's dignity can be just as real.
● Taking down one post also doesn't take down the problem. The same clip gets re-uploaded, re-shared, and mirrored across platforms faster than any single takedown request can keep up with.
● In the meantime, the content simply stays live. Investigations can take hours, sometimes days, and every hour a fabricated video stays up is another hour it keeps spreading.

What Could Actually Fix This ?

A few changes could close most of that gap:

● Content under investigation should be disabled by default, not left visible while a platform decides.
● Once something is confirmed to be synthetically generated, every copy of it should come down, not just the original post, regardless of whether it's obscene.
● The moment a report is filed, the platform should limit the content's reach and turn off sharing, so the damage doesn't outrun the review.

The Bigger Picture

At the end of the day, this isn't just about stopping fake videos from spreading online. It's about whether a person still gets a say in how their own face and voice are used. Right now, that answer depends on how fast a platform decides to act, and that isn't good enough. If these rules are going to mean anything, they need to treat every kind of fabricated content, not just the obscene kind, as something that deserves to come down quickly.

By Rathnika Sidhabatula, 
1/5 BA LLB 
Pendekanti Law College.

Popular posts from this blog

ONE NATION ONE ELECTION: RESHAPING INDIA’S POLITICAL LANDSCAPE || KURELLA HARI KRISHNA

CRYPTOCURRENCY || AFIFA AHMED

Media Trial: Headlines vs Justice in India