AI-generated videos are increasingly being used to create false or misleading narratives in India. Recent fact checks have identified synthetic footage, AI-manipulated videos and cloned audio being circulated as real events, making it harder for social media users to distinguish genuine reporting from fabricated content.
AI-generated videos are becoming a growing misinformation problem
The spread of AI-generated videos in India has become a significant fact-checking concern as generative tools make realistic visual content easier to produce.
Recent cases show that the problem is not limited to completely fictional videos. Artificial intelligence can be used to alter an existing recording, replace a person’s face or clothing, generate a new voice, or create an entirely synthetic scene.
In September 2026, BOOM found that a video claiming to show Ministry of External Affairs spokesperson Randhir Jaiswal responding to a reporter’s question during the BRICS Summit was an AI-manipulated deepfake. The original press conference did not contain the interaction shown in the viral clip, while AI detection tools also flagged the video as manipulated.
The case illustrates why users cannot rely only on a video’s realistic appearance. A convincing video can still contain fabricated elements.
Satya Niketan case shows how AI footage gets recycled
One of the clearest recent examples emerged after the September 6 building collapse in Delhi’s Satya Niketan area.
Two videos were circulated online as footage of the tragedy. BOOM found that neither video actually showed the recent collapse. One video contained AI-generated elements, while the other was authentic footage from 2017 showing the controlled demolition of an unauthorised building in Jasola Vihar.
The second example is particularly important because the footage itself was real. The misinformation came from its context.
Older footage can be extremely persuasive when paired with a current event because viewers may recognise a genuine demolition, protest or disaster scene and assume that the caption accurately explains what they are seeing.
Fact-checkers traced the older footage to reports published in November 2017, establishing that it could not have shown the 2026 Satya Niketan incident.
This is why checking the original date of a video is just as important as determining whether artificial intelligence was used to create it.
AI can manipulate real videos without creating new footage
AI misinformation does not always involve generating an entire video from scratch.
Existing footage can be altered in several ways. A person’s face can be replaced, their clothing can be changed, speech can be generated or their mouth can be manipulated to make it appear that they are saying something they never said.
A recent example investigated by Alt News involved a viral video showing a woman bathing in the Ganga at Haridwar. The circulating version appeared to show her wearing a bikini, but the original video showed her wearing a different outfit. Alt News found that the viral version had been digitally altered.
The original creator had posted the authentic video on Instagram in June 2026. After the manipulated version went viral in September, she publicly stated that the footage had been altered and said she had filed a legal complaint.
The incident demonstrates how AI-assisted editing can change the meaning of genuine footage without replacing the entire recording.
Fake voices can make real news footage misleading
Audio is another area where artificial intelligence is creating new challenges.
A video may contain genuine footage of a public figure while the speech heard over it has been generated or cloned using AI.
BOOM has previously documented cases involving AI voice cloning of public figures, including a video featuring former Finance Minister Nirmala Sitharaman and former RBI Governor Shaktikanta Das. The visuals were real, but AI-generated voice clones were added to make it appear that the two were promoting a fictitious investment project.
This technique can be particularly dangerous because viewers often trust familiar voices. If a person appears on screen and the audio sounds like their voice, audiences may assume the statement is genuine.
That assumption is no longer reliable on its own.
Users should verify major announcements involving investments, government schemes, financial products or public policy through official websites and established news organisations.
Viral AI videos can exploit breaking news
Breaking news provides an ideal environment for misleading AI content because audiences are looking for immediate visual confirmation.
When a major flood, protest, accident, conflict or political event occurs, users often search social media for videos before established news organisations have published detailed reports.
A fabricated video can therefore reach a large audience before anyone has checked it.
BOOM recently found an AI-generated video that was shared as footage of Nepal police rescuing a girl trapped beneath debris after flooding. The rescue incident itself was real, but the video circulated alongside the claim was generated using AI.
The distinction is important. A genuine event can become attached to fabricated evidence.
This makes breaking-news verification especially difficult. Users may see a real news report and then encounter an unrelated AI video that appears to provide visual confirmation.
Political and international events are frequent targets
AI-generated content is also appearing around political and international events.
Alt News recently identified an AI-generated image showing Prime Minister Narendra Modi with world leaders at the 2026 BRICS Summit. The image was circulated as though it were a genuine photograph from the event.
BOOM also found a deepfake video involving the MEA spokesperson in connection with the BRICS Summit. In another recent case, a video from a protest in Tirupati was falsely presented as showing Delhi Police taking action against protesters during the BRICS Summit in the capital.
These examples show that misinformation does not always depend on completely synthetic material. A real event can be combined with fabricated audio, AI-generated visuals or footage from an unrelated location.
The resulting post may look convincing while making a claim that is unsupported by the original evidence.
How Indian users can spot suspicious AI videos
There is no single visual clue that can identify every AI-generated video. Detection tools can help, but users should rely on several forms of verification rather than one test.
Start by checking the source. Ask who uploaded the video first and whether the account has a credible connection to the event.
Next, examine the details. Look for unnatural facial expressions, inconsistent lighting, distorted hands, changing objects, strange text, unusual shadows or movements that do not appear physically consistent.
Audio should also be checked carefully. Artificial voices may have unusual pauses, unnatural pronunciation or inconsistent background sound, although sophisticated voice cloning can avoid obvious warning signs.
Most importantly, compare the claim with independent reporting. If a major incident supposedly involving a senior government official occurred, credible news organisations and official channels would normally provide additional information.
Reverse searches can reveal the original footage
Users can also investigate a suspicious video by extracting screenshots or keyframes and searching for earlier appearances online.
This approach helped identify the old 2017 demolition footage that was falsely presented as the 2026 Satya Niketan collapse. The search connected the circulating footage with older news reports, providing evidence about its actual origin.
The same principle can be used for photographs and screenshots. A viral image may have appeared online months or years before the event described in the accompanying caption.
Checking the date of the earliest available version does not automatically prove that every claim is false, but it can reveal when the material was originally published and whether the current description makes sense.
Users should also remember that social media upload dates are not always the same as recording dates. Reposts, archived material and edited clips can complicate the timeline.
Fact-checking needs more than AI detection tools
AI detection systems can be useful, but they should not be treated as a final authority.
Professional fact-checkers typically combine several methods. These can include reverse image searches, frame-by-frame analysis, source verification, comparison with original broadcasts, checking official statements and examining the history of the content.
In the Satya Niketan case, for example, BOOM combined reverse searching with AI detection tools and comparisons against older news reports.
Alt News has also published recent fact checks involving AI-generated and manipulated content, including visuals connected with protests and viral political claims. The organisation reported that Meta removed several accounts and pieces of content after it flagged AI-manipulated visuals connected with a Jantar Mantar protest.
The wider lesson is simple: no single clue should determine whether a video is genuine.
What users should do before sharing AI content
The safest approach is to pause before sharing a sensational video, especially when it involves a political leader, religious community, government official, celebrity, disaster or security incident.
Ask three basic questions:
Who originally published the footage?
Does independent evidence confirm what the video claims?
Can the video’s date, location and context be established?
If the answer to any of these questions is unclear, the content should not be presented as confirmed information.
Users should also avoid forwarding unverified videos through WhatsApp groups simply because the footage looks convincing. A misleading post can travel rapidly through private messaging networks, where it may be harder for fact-checkers to correct the original claim.
Why AI misinformation matters for Indian audiences
India’s large and highly active social media ecosystem means misleading visual content can reach audiences quickly across languages and regions.
The risk is particularly high for people who rely heavily on short videos for news. A short clip can communicate a powerful impression without providing enough information to establish when or where it was recorded.
AI makes that problem more complicated because the visual evidence itself can now be manufactured or altered.
At the same time, not every unusual video is AI-generated. A strange-looking clip may simply have been compressed, edited or recorded in difficult conditions. Fact-checking therefore requires evidence rather than assumptions.
The growing number of recent Indian cases shows why digital literacy and verification are becoming increasingly important. Before believing or sharing a viral video, users need to check the source, timeline and surrounding evidence.
Takeaways
- AI-generated and AI-manipulated videos are increasingly being circulated as genuine footage in India.
- Recent fact checks have identified fabricated visuals linked to events including the Satya Niketan building collapse and Nepal floods.
- AI can alter existing footage, generate new scenes or clone a person’s voice, making visual verification more difficult.
- Users should check the original source, date, location and independent reporting before sharing suspicious videos.
FAQ
How can I tell if a video is AI-generated?
Look for inconsistent faces, hands, text, lighting, movements and audio, but do not rely on visual clues alone. Check the original source, search for earlier versions of the footage and compare the claim with credible reporting.
Can a real video still be misleading?
Yes. A genuine video can be paired with a false date, location or description. Recent fact checks found 2017 demolition footage being presented as footage of the 2026 Satya Niketan building collapse.
Are AI detection tools always accurate?
No detection tool should be treated as conclusive on its own. Professional fact-checking generally combines detection tools with reverse searches, original footage, credible reporting and other evidence.
What should I do before sharing a suspicious viral video?
Check who originally posted it, when and where it was recorded, and whether independent sources confirm the claim. If those details cannot be established, avoid presenting the video as verified information.
(Internal keyword suggestions: AI-generated videos India, AI deepfake videos India, AI misinformation India, viral AI videos fact check, deepfake fact check India, AI manipulated videos, fake videos on social media, how to identify AI videos, social media misinformation India, AI deepfake news)
