Videos circulating in India show Muslim offenders: a Muslim man snatching a phone in Gwalior, a Muslim woman throwing slippers at a statue of Shivaji Maharaj in Pune
By The Conspiratory EditorsAugust 12, 2026
Where the evidence lands: ContradictedThat viral videos of ordinary crimes and public-order incidents in India routinely document Muslim offenders, and that the religious identity attached to such clips in their captions is reliable information about who was involved.
Contradicted. This file rates a format rather than any single clip, and the format is a lie told with one word. In the two cases checked by the Indian fact-checking organisation Factly on 12 August 2026, the video was real, the incident was real, and the police response was real. Only the religious identification was invented. In Gwalior, police named the man detained after a phone snatching and the station house officer stated he does not belong to the Muslim community. In Pune, the senior police inspector handling a case over slippers thrown at a statue of Shivaji Maharaj stated the woman arrested is not Muslim. Both claims were rated false. Nothing here asserts that any of these people did what they are accused of; both matters are with the police, no court has found anything, and this file takes no view on the underlying incidents. What it rates is the communal label welded on afterwards, and that label is a fabrication with a documented history of preceding mob violence.
Believed by: A large Indian social-media audience on WhatsApp, X and Facebook, reached through short clips of genuine incidents whose captions supply an identification the footage cannot show
Alt News rated the Rajbari video the same day and its account corrects this file's earlier entry in one important respect. The shop owner assaulted in the clip, Zakia Sultana, is Muslim. The viral caption did not merely supply a religion where the footage showed none: it inverted the victim's. A second version circulating alongside it recast her again, as a Dalit teacher, so the same 27 seconds carried two incompatible identities in the same week. The detail worth keeping, though, is what happened next. The X account that posted the clip on 31 August as a Hindu teacher attacked by Islamists went back and corrected itself in a reply, saying she was reportedly assaulted by shopkeepers after complaining about a shop under consumer protection law. The original had been reshared more than three thousand times by the time Alt News wrote it up, and the correction sat underneath it. A retraction from the person who started it, on the same thread, in public, and the false version kept moving anyway. That is a harder problem than a missing label, because here nothing was missing. source →
The format crossed into Bangladesh on 2 September, with the same shape and the same correction. A video of a woman being attacked by a group of men while police shield her circulated with the claim that a Hindu teacher had been assaulted by Islamists. Factly traced it to Rajbari, where the woman is a clothing trader, not a teacher, and where the officer in charge of Rajbari Sadar Police Station said the incident arose from a business dispute with a former employee and had no religious dimension. Bangladeshi outlets covering the same footage report it the same way. So the elements the viral caption supplied, the victim's profession, the religion of everyone involved and the motive, were all added after the fact to real footage of a real assault. Note what that costs: an actual violent incident, with an actual victim, was overwritten to serve an argument about religious persecution that the investigating officer says is not what happened. source →
The format this file rates turned up in Canada the same week, which is worth recording because it bears on whether this is an Indian pattern or simply a pattern. On 31 August Lead Stories rated a claim, posted to X on 27 August, that a 68-year-old Palestinian Muslim migrant had been arrested for burning down a Catholic church. The fire at Saint Simon Church in New Brunswick was real and the steeple did collapse. The man police named, Gilles Mallet, is a local resident, and as of 31 August there was no evidence he is Palestinian, Muslim or a migrant; the Royal Canadian Mounted Police said its investigation found no evidence the fire was religiously motivated or that he has ties to any other country. He has been charged and nothing here speaks to whether he did it, which is a matter for a court. What this file rates is the identity welded on afterwards, and the structure is identical to the Gwalior and Pune cases below: a real event, real footage, real police involvement, and one invented attribute that changes what the event is taken to mean. Two differences are instructive rather than reassuring. The Canadian version supplies three attributes at once, nationality, religion and migration status, where the Indian versions typically add one. And it arrived pre-packaged with flag emoji and a BREAKING label, borrowing the appearance of a wire alert. The correction came the same way it does in India, from a named police force answering a question, four days later. source →
Factly published two checks within two hours of each other on 31 August that together show this genre doing something the rest of the file does not cover. Both videos were scripted from the start. In the first, a man with a tilak stands behind an elderly man in a skullcap making odd gestures, shared as a Hindu harassing a Muslim. Factly traced it to the Instagram account @bhanuchanddubey, where it was posted on 24 August 2026 captioned 'Comedy', one of a run of videos featuring the same two men in different storylines. In another of them the uploader speaks about his co-star and asks viewers not to believe everything they see: 'Our friendship is very deep; he's like my uncle. We make videos showing squabbles, but we live with love.' In the second check, a video shared as a Muslim woman destroying a bike because she disliked the colour saffron was also scripted, and one of the performers, Shivansh Yadav, says in his own Facebook bio that they film such videos for entertainment. Note what separates these from the Gwalior and Pune cases below. There the footage was real and only the religious label was added. Here there is no incident at all: the antagonism itself is performed, by people who are friends or colleagues, and who say so publicly on the accounts the clips came from. The lie is still added downstream in the caption, so the detection method is unchanged, but the thing being relabelled has shifted from a real event to a piece of entertainment. Two details are worth carrying forward. The first is the symmetry: one clip casts a Hindu as the aggressor and the other a Muslim, rated the same day by the same organisation, which is a plain demonstration that the format is indifferent to direction and manufactures communal antagonism rather than reporting it. The second is that neither creator was hiding anything. Both labelled their work as comedy or entertainment on the accounts that published it, which puts the deception entirely in the hands of whoever stripped that label off. As ever, this file rates the caption, not the people in the video, and nobody here did what the posts accuse them of, because there was nothing to do. source →
A second instance six hours after the one below, and this time the source hands us the pattern rather than another example of it. Fact Crescendo reported on 20 August that video of a woman throwing slippers at a statue of Shivaji Maharaj in Pune was circulating with a communal framing identifying her as Muslim. She is not, and the outlet found no communal element in the case at all; it rated the claim Misleading. The video is real and genuinely provocative, which is the whole mechanism: the outrage is generated by authentic footage and the false attribution is fitted to it afterwards, exactly as it was in the Surat murder logged this morning. What is worth adding is that this is not a run of coincidences. Fact Crescendo maintains a standing series of these checks, including a Manipur assault video in July 2024, a Bangladeshi clip recaptioned in June 2026 as a Muslim girl being forced to chant, a man lighting a cigarette from a temple lamp in July 2026, and a Kanpur woman attacked in a family dispute the same month, each debunked on the same finding that no communal element existed. This file has argued that the misattribution is a format rather than a series of accidents. One newsroom's archive of it, spanning two years and four states, is the evidence for that claim rather than a restatement of it. source →
Another instance, and this one is useful because the rebuttal came from the investigating police rather than from inference. On 19 August Factly examined a viral screenshot claiming that a 39-year-old man killed in the Bhestan area of Surat on 5 August had been murdered by Muslim youths. The officer in charge of Bhestan Police Station told Factly directly that all nine people arrested, four of them minors, are Hindus, and that the killing followed a long-running personal dispute. Reverse image search led to multiple August 2026 news reports carrying the same visuals, none of which mention any communal element. The accused are not named here and neither is the other party to the dispute, since none of that is the point. What matters for this file is the sequence, which is the same one it has documented repeatedly: a real, violent, locally reported crime supplies the attention, and the communal attribution is fitted to it afterwards by people with no connection to the case. The victim was real and so was the killing. Only the motive was invented, and it was invented by strangers. source →
A variant that works by subtraction rather than by smear, and it is worth distinguishing from the rest of this file. Factly reported on 17 August that a video circulating as footage of Muslims in Afghanistan hoisting the Indian flag and singing the Indian national anthem is genuine footage from India: it was recorded at Darul Uloom Deoband, the Islamic seminary in Saharanpur, Uttar Pradesh, on Republic Day, 26 January 2026, and traced through a reverse image search to a post of 27 January and to media coverage of the same event. Note what the false caption actually does. The usual move catalogued here takes real footage and relabels the people in it to make a community look guilty of something. This one takes real footage of an Indian Muslim seminary honouring the Indian flag and relocates it to another country, which does not accuse anyone of anything; it simply removes the fact that Indian Muslims were doing it. The claim spread in the week of India's 80th Independence Day, alongside a separate Factly check on old BJP training footage recirculated as party leaders hoisting a party flag rather than the tricolour, so the same technique ran in both political directions within days. source →
Opening entry, built off two Factly checks published within five hours of each other on 12 August. The pairing is the argument. One clip is from Gwalior in Madhya Pradesh and shows a phone snatching on 8 August; the other is from a Pune suburb and shows slippers thrown at a statue on 10 August. Different states, unrelated incidents, no shared footage, and the same single fabrication in each: a religious identity the camera never recorded. Both were settled the same way, by an identified police officer stating on the record who the person actually is. If further instances are checked, they belong here rather than in new files, because the instance is never the story. source →
The full story
Two clips, two states, one edit
On 8 August in Gwalior, Madhya Pradesh, two sisters riding a scooter had a phone snatched by a man on a motorcycle. They chased him, caught him, and handed him to the police, who filed a report. Somebody filmed it.
On 10 August in Sasane Nagar, a suburb of Pune in Maharashtra, a woman threw slippers at a statue of Chhatrapati Shivaji Maharaj and was detained. Footage shows her surrounded by a crowd with ink on her face and female officers present. Kalepadal Police Station filed a case and made an arrest that day, and said it was investigating why.
Both incidents happened. Both videos are genuine. Both produced real police cases. And as each clip travelled, one sentence was added: the man was a Muslim, the woman was a Muslim.
On 12 August the Indian fact-checking organisation Factly published on both, hours apart, and rated both false. That is what this file is about. Not either clip, but the operation they have in common.
How each one was actually settled
The method matters more than the verdicts, because it is not the method this site usually documents.
For Gwalior, Factly ran reverse image searches on keyframes, which led straight to legitimate news coverage of the 8 August robbery. So far this is ordinary verification, and it confirms the video rather than undermining it. Then the organisation telephoned Padav Police Station. The station house officer, Shailendra Bhargava, said the man detained is Manpreet Singh, son of Jaspal Singh, a native of Punjab, and that he does not belong to the Muslim community.
For Pune, the same sequence. Keyframe searches led to media posts carrying the same visuals from 10 August. Factly then contacted Kalepadal Police Station, where Senior Police Inspector Girisha Nimbalkar said the woman arrested is Ashwini Dagde and is not Muslim, and confirmed the case and the arrest.
Both corrections rest on a named officer, speaking about a file their own station is handling. Nothing about either identification is a deduction or an inference. It is the police saying who they arrested.
One thing this file does not do, and will not: it takes no position on what either person did. Both matters are with the police, neither has been before a court, and the presumption of innocence applies to both regardless of what any caption says about them.
Why every detection method comes back clean
Almost everything else in this archive can be caught by looking harder at the file. A video is older than claimed, or from another country, or generated. Reverse image search finds the original. Detectors flag synthetic audio. Watermarks survive. Metadata disagrees with the caption.
None of that reaches this. The footage was not edited, so there is nothing to detect. It is not old, so the date checks out. It is not from elsewhere, so the location checks out. Every instrument returns green, and the false statement sits entirely outside the frame, in a line of text typed next to it.
The only tool that works is a reporter with a telephone and a police station number, which is slow, local, and does not scale. It also depends on an officer being willing to answer, which they were in both of these cases and may not be in the next. That asymmetry, one sentence to make the claim and two phone calls across two states to unmake it, is the whole economics of this format.
What the added word is for
It is worth being plain about the function of the edit, because it is not decorative.
A phone snatching in Gwalior is a matter for one police station and two sisters who got their phone back. Attach a community to it and the same thirty seconds of footage becomes a claim about millions of people, distributed nationally, arriving with what looks like evidence already attached. The clip does the persuading; the word does the aiming.
That is a false and communally targeted claim, and this file names it as one. India has a documented history of rumours attaching criminality to a group preceding mob attacks, which is the same dynamic our files on child-snatching panics and lynching rumours trace, and it is why Indian fact-checkers treat this format as urgent rather than as a caption error. The harm does not require the audience to remember the individual case. It requires only that the impression accumulate.
Why it keeps working on careful people
The uncomfortable part is that this format defeats diligence rather than rewarding credulity. A person who does the recommended thing, checking whether the video is real, comes away more confident, not less, because it is real. Verification confirms the container and never touches the contents.
The claim also arrives in the grammatical form of a description rather than an argument. Captions identify people; that is what captions are for. A sentence saying who is in a video does not announce itself as a contention to be weighed, so it is filed as information rather than as a proposition.
And it accumulates. Somebody who has seen this caption on twenty clips over two years experiences the twenty-first as confirmation of a pattern they have observed themselves. The pattern is real, but it is a pattern in the captions rather than in the incidents, and from the inside those are indistinguishable.
One word, two phone calls
Two videos this week showed two real incidents, filmed accurately, handled by two police forces who filed cases and made arrests. In both, the single false element was a religious identification that no camera recorded and no viewer could have checked.
It took a named station house officer in Gwalior and a named senior inspector in Pune, answering questions about their own case files, to establish that the identifications were wrong. That is what it costs to remove one word, and the word costs nothing to add.
The practical lesson is narrow and worth holding: when a clip of a real incident comes with a community attached, the video being genuine tells you nothing at all about whether that part is. Those are two separate claims, and only one of them can be checked by watching.
What's still unexplained
- How much of this format is coordinated and how much is individual reflex is not established by either check. Two instances in one day show the shape is widespread; they do not show who is typing the captions or why.
- Whether the corrections reach the audiences that saw the originals is undocumented and, on the evidence of how such things usually move, unlikely. Neither check reports circulation figures for the false version or the correction.
- Whether police statements are a durable remedy is unclear. They worked in both August cases because officers answered and spoke on the record, and there is no reason to assume that will hold in every case, or that a denial reaches the people who saw the accusation.
Point by point
The claim: The video is genuine, so the caption describing who is in it must be reliable too.
What the record shows: These are separate claims and only the first one can be checked by looking. In both August cases the footage was authentic and the incidents happened as filmed, which is precisely what makes the format effective: a viewer who verifies the clip finds it real and stops. Religion is not visible in a scooter chase or in a crowd around a statue. The caption is supplying information that the camera never recorded, and it is the only part that was invented.
The claim: Fact-checkers cannot really disprove someone's religion, so these debunks are just assertion.
What the record shows: They are not assertions, and how they were established is the most useful thing here. In Gwalior, Factly contacted Padav Police Station and the station house officer, Shailendra Bhargava, gave the detained man's name and parentage and said he is not from the Muslim community. In Pune, Senior Police Inspector Girisha Nimbalkar named the woman arrested and said she is not Muslim. Both are identified officers speaking on the record about cases their stations are handling, which is the strongest form of identification available short of a court record.
The claim: This is an isolated mistake rather than a pattern.
What the record shows: Two independent instances, in Madhya Pradesh and Maharashtra, on unrelated incidents four days apart, both checked and both false on the same day, is not a coincidence of captioning. The same operation runs on each: take a real clip of a real offence, leave the footage untouched, and add a religious identity. The Indian fact-checking ecosystem processes this format continuously, which is why two arrived in a single day's output.
The claim: The standard verification toolkit will catch this.
What the record shows: It will not, and that is the technical point worth carrying away. Reverse image search establishes that a clip is old, or from elsewhere, or synthetic. None of those apply here. AI detectors return nothing because nothing was generated. Metadata is intact because the file was not manipulated. Every method this site normally documents comes back clean, because the fabrication is not in the video. It is in a sentence typed beside it, and the only instrument that reaches it is a reporter telephoning the officer who has the file.
The claim: Correcting the identity is pedantic when a crime clearly took place.
What the record shows: The identity is the entire payload. A phone snatching in Gwalior is a local matter for the Padav police; the same clip captioned to name a community is an argument about that community, distributed nationally. That conversion is the purpose of the edit, and its consequences are documented: rumors attaching criminality to a group have preceded mob attacks in India often enough that fact-checkers treat this format as urgent rather than trivial. On the incidents themselves this file says nothing. Both are with the police, neither has been to court, and everyone involved keeps the presumption of innocence.
Timeline
- 2026-08-08In Gwalior, Madhya Pradesh, two sisters riding a scooter, named in local reporting as Harshita and Rishita Agarwal, are robbed of a phone by a man on a motorcycle. They give chase, catch him, and hand him to the local police. A first information report is filed. The incident is covered by several news outlets and the footage circulates.
- 2026-08-10In Sasane Nagar, Pune, Maharashtra, a woman throws slippers at a statue of Chhatrapati Shivaji Maharaj and is detained. Footage shows her surrounded by people, with ink smeared on her face, and with female police officers present. Kalepadal Police Station files a case the same day and makes an arrest, and says it is investigating why she did it.
- 2026-08 (both cases)As each clip spreads, the captions add an identification the footage cannot support. The Gwalior video is shared as two women thrashing a Muslim man who snatched their phone. The Pune video is shared as a Muslim woman throwing slippers at the statue. Nothing visible in either clip establishes anybody's religion.
- 2026-08-12Factly publishes on the Gwalior clip. Reverse image searches on keyframes lead to the original news coverage of the 8 August incident. The organisation contacts Padav Police Station, and the station house officer, Shailendra Bhargava, states that the man detained is Manpreet Singh, son of Jaspal Singh, a native of Punjab, and that he does not belong to the Muslim community. The claim is rated false.
- 2026-08-12Factly publishes on the Pune clip the same day. Keyframe searches lead to media posts carrying the same visuals from 10 August. The organisation contacts Kalepadal Police Station, and Senior Police Inspector Girisha Nimbalkar states that the woman involved is Ashwini Dagde and is not Muslim, and confirms a case was filed and an arrest made on 10 August. The claim is rated false.
- 2026-08-12Two checks on one day, in two states, on two unrelated incidents, running the identical operation. That is the reason this file exists: not either clip on its own, but the format they share.
- Longer runThe format has a documented history in India, and it belongs to the same family as the rumor-driven mob violence covered in our files on child-snatching panics and lynching rumors. A false communal identification attached to a real, filmed incident is one of the more dangerous artifacts in circulation, because it arrives with proof of the incident already attached.
Contradicted. This file rates a format rather than any single clip, and the format is a lie told with one word. In the two cases checked by the Indian fact-checking organisation Factly on 12 August 2026, the video was real, the incident was real, and the police response was real. Only the religious identification was invented. In Gwalior, police named the man detained after a phone snatching and the station house officer stated he does not belong to the Muslim community. In Pune, the senior police inspector handling a case over slippers thrown at a statue of Shivaji Maharaj stated the woman arrested is not Muslim. Both claims were rated false. Nothing here asserts that any of these people did what they are accused of; both matters are with the police, no court has found anything, and this file takes no view on the underlying incidents. What it rates is the communal label welded on afterwards, and that label is a fabrication with a documented history of preceding mob violence.
Reviewed by The Conspiratory Editors · Last reviewed September 2, 2026 · How we rate
Common questions
Is The added word true?
Contradicted. This file rates a format rather than any single clip, and the format is a lie told with one word. In the two cases checked by the Indian fact-checking organisation Factly on 12 August 2026, the video was real, the incident was real, and the police response was real. Only the religious identification was invented. In Gwalior, police named the man detained after a phone snatching and the station house officer stated he does not belong to the Muslim community. In Pune, the senior police inspector handling a case over slippers thrown at a statue of Shivaji Maharaj stated the woman arrested is not Muslim. Both claims were rated false. Nothing here asserts that any of these people did what they are accused of; both matters are with the police, no court has found anything, and this file takes no view on the underlying incidents. What it rates is the communal label welded on afterwards, and that label is a fabrication with a documented history of preceding mob violence.
What is The added word?
Two videos went around India this week. In one, two sisters on a scooter in Gwalior chase down and catch a man who has snatched their phone. In the other, a woman throws slippers at a statue of Shivaji Maharaj in a Pune suburb and is detained. Both happened. Both were filmed. Both produced police cases. The captions added one thing each:…
What does the evidence show?
These are separate claims and only the first one can be checked by looking. In both August cases the footage was authentic and the incidents happened as filmed, which is precisely what makes the format effective: a viewer who verifies the clip finds it real and stops. Religion is not visible in a scooter chase or in a crowd around a stat…
Why do people believe it?
Everything checkable checks out. The video is real, the location is real, the arrest is real, and a person who verifies those things has done more than most and still ends up with the false part intact.
What is still unresolved?
How much of this format is coordinated and how much is individual reflex is not established by either check. Two instances in one day show the shape is widespread; they do not show who is typing the captions or why.
Sources
- 1.An old video of the Indian flag hoisting at Darul Uloom Deoband is falsely shared as Muslims hoisting the flag in Afghanistan, Factly (2026)
- 2.The phone snatcher caught by sisters in Gwalior is not a Muslim, Factly (2026)
- 3.The woman who threw slippers at the Shivaji Maharaj statue in Sasane Nagar, Pune, is not Muslim, Factly (2026)
- 4.Factly (International Fact-Checking Network signatory), Factly (2026)
- 5.Shivaji (background on Chhatrapati Shivaji Maharaj and his standing in Maharashtra), Wikipedia (2026)
- 6.Gwalior, Wikipedia (2026)
- 7.Misinformation in India (background on the WhatsApp-driven rumor environment), Wikipedia (2026)
Embed this verdict on your site
Paste this snippet to show our sourced verdict as a small card, with a link back to the full case file. Free to use.
<iframe src="https://theconspiratory.com/embed/communal-misidentification-india" title="The Conspiratory verdict" width="520" height="190" style="border:0;max-width:100%" loading="lazy"></iframe>Help us investigate
This is a living case file. If you spot an error or know evidence we missed, tell us, and weigh in on where you land.
Where do you land?
Cast your read on this one.
Comments
Add your take. Comments are read and approved by a human before they appear, so keep it on topic and civil. Please do not accuse named, living people of crimes.