OOmnistu

Blog / Security

Security

The video that was never really them

The I4C handbook explains how deepfakes spread, and India's amended IT rules now require platforms that enable creating or sharing realistic synthetic media to label it, and large social platforms to check uploaders' declarations. A label carries information; a missing label does not make a video genuine. The checks that matter are verification through trusted sources before believing or forwarding.

OOmnistu
Reviewed 2026-11-21 · 4 min read

A clip arrives in a family group: a familiar face, a familiar voice, saying something that asks you to act. Two Indian official documents are useful here from opposite directions. The Indian Cyber Crime Coordination Centre's handbook of scam dos and don'ts explains what a deepfake is and how it travels. And the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, 2026 - notified on 10 February 2026 and in force from 20 February 2026 - created a labeling regime for synthetically generated media. Between them sits a practical answer: what a label can and cannot tell you, and the checks that still rest with you.

What a deepfake is

The handbook's description is plain: cybercriminals use advanced AI to create fake videos or audio clips by manipulating real footage or recordings. These fake media are then spread through social media, messaging apps and emails, often targeting public figures, celebrities or people in authority, with the goal of deceiving viewers, manipulating opinions or spreading false information. The handbook adds that criminals may use social engineering techniques to make the deepfake seem real - the clip rarely arrives alone; it arrives with a story that explains why you should believe it and why you should act now.

What the amended rules require

The 2026 amendment writes a definition into the rules: synthetically generated information is audio, visual or audio-visual information created, generated, modified or altered using a computer resource so that it appears real, authentic or true and depicts a person or event in a way likely to be perceived as indistinguishable from a natural person or real-world event - with carve-outs for routine good-faith editing, formatting and accessibility improvements, and for documents and training material that do not create false records. On top of that definition, the rules place obligations on intermediaries. A platform that offers a service enabling the creation, sharing or dissemination of such content must take measures against unlawful synthetic content - including content that falsely portrays a real person in a manner likely to deceive - and must prominently label other synthetic content: a clearly visible label for visuals, a prominently prefixed disclosure for audio, plus permanent metadata or another provenance mechanism, where technically feasible, with a unique identifier. The platform may not enable modification, suppression or removal of that label. Significant social media intermediaries must additionally ask users to declare synthetic content before publishing, deploy technical measures to verify those declarations, and display a clear and prominent label when the declaration or verification confirms the content is synthetic.

What a label can and cannot tell you

These are obligations placed on platforms, and they matter - but their limits are built into the design. A label carries real information: the uploader declared the content synthetic, or the platform's checks confirmed it. A missing label, however, does not make a clip genuine. Declarations can be false, technical verification has limits, routine editing is exempt, older clips predate the regime, and media forwarded through chat apps can arrive stripped of whatever context the original post carried. The reverse also holds: a label is not an accusation. Synthetic media has legitimate uses, and the rules' own carve-outs exist because not every altered clip is a deception. So neither the presence nor the absence of a label settles the question that actually matters - whether what you are watching is what it claims to be.

The checks that still rest with you

The handbook's habits are built for exactly that gap. Verify content: always check the authenticity of media before sharing or believing it. Use trusted sources: rely on reputable platforms for news and updates. Do not share unverified media, do not trust suspicious sources, and be cautious of content that seems exaggerated or emotional - urgency and outrage are the social engineering the handbook warns about. If a clip appears to show someone you know asking for money, gift cards or credentials, treat the request as unverified until you have confirmed it through a channel you already know is theirs - a call to their usual number, not a reply to the clip. The handbook also advises reviewing privacy settings and limiting the personal information you share online, since public photos, videos and voice notes are the raw material fakes are built from. And it asks readers to report suspicious content to the platform or the authorities.

If a fake has already cost you

If money has moved from your accounts because of a fake video or voice, call 1930, the national financial-fraud helpline, immediately. Report the incident at cybercrime.gov.in, the national portal, which is also the route when nothing was lost. Prompt reporting can help the authorities intervene, but it does not assure prevention or recovery.

The shape worth remembering

India's labeling rules give you one more signal, and the I4C handbook tells you how much weight to put on it: some, but never all of it. A label says a clip was declared or confirmed to be synthetic; no label says nothing at all. The question the handbook keeps returning to is the one worth keeping: has this been verified through a source you independently trust - before you believe it, and before anyone else receives it from you.

Sources: the I4C handbook of cyber-scam dos and don'ts (Indian Cyber Crime Coordination Centre, Ministry of Home Affairs; PDF created March 2025; hosted by NDMC) for the description of deepfakes, how they spread, and the verification, sharing and privacy habits, and the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, 2026 (G.S.R. 120(E), notified 10 February 2026, in force 20 February 2026) for the definition of synthetically generated information and the rule 3(3) labeling, metadata and label-protection obligations and the rule 4(1A) declaration and verification obligations. Fraud reporting: 1930 and cybercrime.gov.in.

Frequently asked questions

The clip has no AI label. Does that mean it is real?

No. Under the IT Amendment Rules, 2026, the declaration-and-verification duty sits with significant social media intermediaries, while platforms whose services enable creating or sharing synthetic media carry the labeling duty; the rules also exempt routine good-faith editing from the definition; a clip can also reach you through a forward that stripped its original context. Absence of a label is not evidence of authenticity - it only means no label was applied. The I4C handbook's standing advice covers this case directly: check the authenticity of media before believing or sharing it, and rely on trusted sources for confirmation.

The video is labelled AI-generated. Is it a scam?

Not by that fact alone. A label means the content was declared or confirmed to be synthetically generated - and synthetic media has legitimate uses - satire, film effects, accessibility tools - which the rules regulate through labeling rather than prohibition. Separately, the definition itself excludes routine good-faith editing, accessibility improvements and similar non-deceptive processing, so such material is not treated as synthetic at all; a satirical or film clip that does look real would still be labelled, not exempted. What matters is what the clip asks of you: if it solicits money, credentials or onward sharing on the strength of someone else's identity, treat that request as unverified whatever the label says, and confirm through a channel you independently trust.

A video or voice note that sounds like a family member is asking for money. What do I do?

Do not act on the clip itself. The I4C handbook describes fake videos and audio built from real recordings and spread through messaging apps, made convincing with social engineering - so sounding like them is not proof it is them. Confirm the request through a channel you already know: call the person on their usual number, or ask in person. If money has already moved, call 1930 immediately and report at cybercrime.gov.in; prompt reporting can help the authorities intervene, but it does not assure prevention or recovery.

I think I have found a deepfake. Should I report it?

Yes - the handbook asks readers to alert the platform or the authorities when they find potential deepfakes, and to avoid forwarding the content while doing so, since spreading unverified media is how the pattern grows. You can report through the platform's own reporting tools and at cybercrime.gov.in. Under the amended IT rules, platforms also have obligations around synthetic content, including labeling confirmed synthetic media, so a report gives the platform the awareness those obligations attach to.