AI-generated videos are becoming increasingly realistic, making it harder for Indian internet users to distinguish genuine footage from manipulated content. Recent deepfake scams involving public figures show why checking the source, context and authenticity of viral videos matters before believing or sharing them.
Why AI-generated videos are becoming harder to spot
AI-generated videos have moved well beyond obvious face-swapping clips with unnatural expressions. Modern video-generation systems can create realistic people, environments, movement, voices and background sounds, making visual inspection alone less reliable.
Recent advances in AI video generation have contributed to this shift. Google’s Veo 3.1, for example, can generate vertical videos and improve the consistency and realism of generated scenes. Other widely available AI video tools have similarly lowered the technical barrier for creating convincing synthetic footage.
This matters in India because social media platforms are a major source of information for millions of users. A convincing video can spread rapidly through Instagram, Facebook, YouTube, WhatsApp and other messaging platforms before anyone has established whether it is authentic.
The problem is no longer simply identifying a video that looks artificial. Users increasingly need to verify where a video came from, whether credible sources have reported the same event and whether the footage has signs of manipulation.
Deepfake scams have already affected Indian users
The risks are not theoretical.
In August 2026, The Indian Express reported cases in Pune involving deepfake videos of prominent public figures. In one case, a woman was allegedly defrauded of more than Rs 8 lakh after encountering a fake video featuring Union Finance Minister Nirmala Sitharaman that promoted an investment scheme. The victim reportedly shared personal information and later transferred money to accounts controlled by fraudsters.
Another Pune case involved a nurse who was allegedly duped of Rs 4 lakh after seeing a deepfake video of the late industrialist Ratan Tata promoting an investment opportunity on Instagram.
These cases demonstrate an important point: a familiar face does not make a video authentic.
Fraudsters can combine an AI-generated face or voice with investment claims, fake advertisements and links designed to collect personal information. That means users should treat unexpected financial advice or investment offers in viral videos with particular caution.
1. Check the original source before trusting the video
The first step is to find out where the video actually came from.
A video shared by a WhatsApp contact or reposted by an Instagram account is not necessarily the original recording. Look for the earliest available upload and check whether the account publishing it is credible.
If a video claims that a minister, actor, government department or major company made an announcement, search for the same information through the organisation’s official website or verified social media accounts.
For example, if a viral clip claims that a Union minister announced a new government investment scheme, do not rely on the video itself. Check the relevant ministry’s official communication and established news organisations.
The Press Information Bureau’s Fact Check Unit is another useful resource for claims involving the Central Government. The government said in April 2026 that its fact-checking unit had published more than 2,900 fact checks and had identified fake claims involving deepfakes, AI-generated videos, misleading notifications and websites.
The key rule is simple: verify the claim independently rather than allowing the viral video to serve as its own evidence.
2. Look for visual and audio inconsistencies
Human eyes are still useful, but they should be treated as an initial screening tool rather than final proof.
When watching a suspicious video, look closely at the person’s mouth movements, facial expressions, hands and interactions with objects. Pay attention to sudden changes in lighting, unusual reflections, distorted text or objects that appear to change shape between frames.
Audio can also provide clues. Listen for unnatural pauses, strange pronunciation, inconsistent background noise or a voice that does not quite match the person’s normal speaking pattern.
However, these clues are becoming less dependable as AI systems improve. A video without obvious glitches can still be manipulated.
The Indian Express recently highlighted that AI video generators can now create highly realistic footage, making traditional visual clues increasingly difficult to rely on.
So, if you do notice something strange, treat it as a warning sign. If you notice nothing strange, do not automatically conclude that the video is real.
3. Search for independent reports of the same event
One of the strongest verification methods is to step outside the original social media post.
Suppose a video claims that an earthquake, major accident, political announcement or celebrity incident happened in a particular Indian city. Search the key details separately using multiple credible news sources.
Check the location, date, people involved and claimed event.
If the incident supposedly happened in Jaipur but there is no credible reporting, official statement or local evidence, that absence should make you cautious. It does not automatically prove that the video is fake, but it means the claim needs more verification.
For major public events, compare reports from established news organisations with official sources such as government departments, police authorities or district administrations.
This is particularly useful for regional audiences. A viral video claiming to show an incident in Nagpur, Indore, Pune, Jaipur or another Tier-2 city should be checked against local reporting rather than accepted because it appears geographically familiar.
4. Check for AI labels, watermarks and content credentials
Some AI-generated media now carries technical signals that can help establish its origin.
Google’s SynthID system, for instance, embeds an imperceptible watermark into content generated by supported Google AI systems. Google says SynthID has expanded beyond images to include AI-generated video, audio and text.
Google also allows users to upload certain videos to Gemini and ask whether they were generated using Google AI. The system can scan for SynthID and indicate where the watermark is detected in the audio or visual content.
Content provenance standards such as C2PA Content Credentials are also being developed to provide information about how digital media was created or edited.
These tools can be useful, but there is an important limitation. A detector designed to identify content from a particular AI system cannot establish that every video without its watermark is genuine.
In other words, a positive detection can be meaningful, while the absence of a particular watermark should not automatically be interpreted as proof of authenticity.
5. Do not share until the claim is independently verified
The final step is also the simplest: pause before forwarding.
This matters especially when a video triggers a strong emotional reaction. Claims involving communal tensions, elections, disasters, celebrity controversies, government announcements or investment opportunities can spread rapidly because people feel compelled to warn others or act immediately.
That urgency is exactly what makes misinformation effective.
India’s government has recognised deepfakes and synthetically generated information as a growing digital safety issue. In March 2026, the Ministry of Electronics and Information Technology said the IT Rules had been strengthened to address harms arising from synthetically generated information, including deepfakes and AI-generated content. The framework includes requirements concerning labelling and traceable metadata for permissible AI-generated content.
The government also said in July 2026 that 13 responsible AI projects had been approved for deepfake detection and that more than 11.37 lakh people had benefited from cyber-security awareness workshops.
Technology and regulation can help, but users remain the last line of defence before misleading content reaches another audience.
Why public figures are frequent targets
Deepfake creators often use recognisable people because familiarity can make viewers lower their guard.
A minister appearing to announce an investment programme may seem credible to someone who recognises the face and voice. Similarly, a celebrity appearing to endorse a product can make an advertisement look legitimate.
Recent cases involving Nirmala Sitharaman and Ratan Tata demonstrate how this technique can be used in financial fraud. The scammers did not need victims to understand how AI worked. They only needed the video to appear convincing long enough for viewers to follow the instructions attached to it.
This is why verification should focus on the claim rather than the appearance.
Ask what the video is asking you to believe or do. If it asks for money, personal documents, passwords, OTPs or financial details, stop and verify through an independent official source.
What Indian internet users should remember
There is no single visual trick that can identify every AI-generated video.
AI detection technology is developing alongside generation technology, while watermarking and provenance systems are becoming more common. But not every video will carry a detectable watermark, and automated detection tools can have limitations.
The safest approach is therefore a combination of methods: check the source, examine the content, search for independent reporting, use available verification tools and avoid sharing until the claim has been established.
The goal is not to become an expert in artificial intelligence. It is to develop a basic habit of digital verification.
A few seconds spent checking a viral video can prevent misinformation from spreading and, in cases involving scams, can protect users from losing money or exposing sensitive personal information.
Key Takeaways
- AI-generated videos are becoming more realistic, so visual clues alone are no longer enough to establish authenticity.
- Check the original source and compare important claims with official statements and credible news reports.
- Tools such as Google’s SynthID can help identify content generated by supported Google AI systems, but a missing watermark does not prove that a video is genuine.
- Never send money or personal information based solely on a viral video, especially when it involves investment schemes or public figures.
FAQs
How can I tell if a video is AI-generated?
Look for unusual mouth movements, distorted hands, inconsistent lighting, unnatural audio or objects changing between frames. However, these clues are not definitive. Source verification, independent reporting and available content-provenance tools provide stronger evidence.
Can Google Gemini detect AI-generated videos?
Gemini can check certain uploaded videos for Google’s SynthID watermark and provide information about detected AI-generated segments. This is primarily useful for content created using supported Google AI systems and should not be treated as a universal AI-video detector.
What should I do if I receive a suspicious deepfake on WhatsApp?
Do not forward it immediately. Check the original source, search for the claim through credible news organisations and official government or institutional accounts, and look for fact checks. If the video involves money or personal information, do not follow its instructions until independently verified.
Can an AI-generated video be completely realistic?
Yes. Modern AI video-generation systems can produce increasingly realistic visuals, movement and audio. This is why relying only on visible glitches is becoming less effective and why checking the source and context is increasingly important.
(Internal keywords: AI-generated videos India, deepfake videos India, how to identify deepfake, AI video verification, fake videos on social media, deepfake scams India, AI misinformation India, how to verify viral videos, SynthID India, AI-generated content detection, deepfake fraud, social media misinformation)
Leave a comment