#FactCheck - Viral Video Claiming Attack on Burj Khalifa is AI-Generated
Executive Summary
Amid rising tensions between the United States, Israel, and Iran, a video is circulating on social media claiming that Burj Khalifa in Dubai has been attacked. The clip is being widely shared with users alleging that a strike took place near the iconic skyscraper. However, research by CyberPeace found the claim to be misleading. Our research revealed that the viral video is not real and was generated using artificial intelligence.
Claim
On March 1, 2026, a Facebook user shared the viral clip claiming that an attack had taken place in Dubai. The post was shared with the caption: “Dubai has been attacked.” The link to the post and its archive is provided below along with a screenshot.

Fact Check
To verify the claim, we first searched Google using relevant keywords. During this process, we found a report published on March 1, 2026, by the Indian news outlet Dainik Bhaskar.

According to the report, tensions in the Middle East escalated amid the Israel–Iran conflict, impacting several countries in the region. A drone incident reportedly occurred near Burj Khalifa, prompting authorities to evacuate the building as a precautionary measure and temporarily switch off its lights. However, the visuals seen in the viral video do not match the details or imagery described in the report. Upon closely examining the viral clip, we noticed several technical inconsistencies and unusual visual elements, raising suspicions that the video might have been generated using artificial intelligence. To verify this, we analyzed the video using the AI detection tool Sightengine. The results indicated a 99% probability that the video was AI-generated.

Conclusion
Our research found that the viral video circulating on social media is not authentic. The footage was created using artificial intelligence and does not depict a real attack on Burj Khalifa.
Related Blogs
.webp)
Introduction
A mass of people gathered in public is a common scene. All are eager for the same information, official estimates are delayed, and the space is then filled by whatever is earliest. In Delhi the gap was partially filled by old videos being described as new footage and by reports of injuries and deaths which police and fact-checkers have rejected. This report investigates what was claimed, by whom, and what has been validated. It does not take sides in the protests. It is focused on a small issue: how this false information travelled, why it was effective, and what the reader can do to combat it.
What was reported
India Today claimed that the ISI, the spy agency of Pakistan, posted a false injury report of a senior opposition leader, as well as fake reports of deaths, during the Delhi protests. That is what the publication's own headline says, and the claim of ISI links is based on sources, not an established fact.
Delhi Police has filed at least four FIRs against social media handles for creating fake content and attempts at building an "anti-India narrative", according to a report by The Week, quoting news agency ANI. Police said multiple Pakistani handlers reported deaths in the 10 October protest and shared fake videos while impersonating Indians. India TV reported that around 500 posts and 100 handles had been removed, citing unidentified sources and the police. There is no independent verification of those numbers in the material we reviewed.
What has been verified, and what has not
The strongest piece is from News Meter, a fact-checking organisation. It found two videos posted on X on 12 October. The first was purportedly a video of police firing on a rally and killing five workers. When searching the video for the beginning of its footage, NewsMeter found it on Facebook from 5 March 2020 and in a Times of India report about the February 2020 riots in Delhi's Chand Bagh locality. The second video, which was reportedly of soldiers beating a protester, has been posted on YouTube since 2019 and on news websites reports as early as 2017, according to News Meter. There is no evidence of firing at the rally. Another of the original fact checkers, NewsMeter, traced a video alleging an attack on a politician's residence to December 2019 rallies. The Delhi Police also labelled the claim of firing at the rally as a rumour on 12 October.
A few items still require substantiation. The ISI's direction is unconfirmed, and the accounts' source is police and other sources. News Meter identified the location data of one account as "South Asia" with a slight possibility of VPN use, and it examined its earlier report of Pakistan-based propaganda accounts' pattern, but nothing is definitive. Location tags are weak evidence. The death claims were deemed false by police, but no external investigation of either accusation exists in the information looked at.
How fabricated visuals travel
Recycled footage is effective because it is footage. The crowd, the violence – all of that is real; they just got the date and place wrong. Writer Claire Wardle and researcher Hossein Derakhshan have dubbed this entirely different genre of falsehood as a form of "misinformation" – where the lie is in the caption, but the pixels still tell the truth.
That's where the urgency comes in. The article published by News Meter closed with the plea to 'people to come on the road and block the city'. Short, scary and sensational messages are among the easiest to share. A 2018 Science paper by Soroush Vosoughi, Deb Roy and Sinan Aral found false news stories are about 70 per cent more likely to be retweeted than accurate stories and spread more rapidly, primarily because humans (not bots) retweeted them.
And platform design does, too. Recommendation algorithms tend to incentivise content engagement, and there's a tonne of engagement to be driven from anger. The credibility of accounts using pseudonyms or pretending to be local seems to come from the community they claim.
Why injury and death claims are different
The most powerful piece of rumour is the one in which a person has been killed or injured. It stirs mourning and fury, inflames crowds to violence, and leaves the rumour-monger little room for denial. Even in cases of retraction, the original story is often the one to stay. Delhi Police in its post cautioned that "malicious rumours" are an offence.
If false reports drown out the conversation, true reports of harm are then easier to write off. Official messaging needs the same due care as a viral post for the same reason: Public officials are party to the events they describe. Tone is no guarantee, not even "certainty". Multiple independent tellings can demonstrate it.
The damage extends further than the audience as well. If people are searching for loved ones, they might find a rumour; before they get confirmation from the individual, helplines can become overwhelmed, and reporters have all their time wasted following up claims that were planned to be followed up.
What responsible sharing looks like
But most of it is a run-of-the-mill habit. Put the brakes on before you forward something that fuels your fear or anger. Investigate who created the post, when the account was created. Reverse image search the videos with Google Lens or Tin Eye (that's how the Delhi videos were dated). Find the same report on outlets that credit their sources. Choose fact checkers that follow clear standards, like signatories of the International Fact-Checking Network code of principles.
If you've already posted something false, admit it and include the correction. Even quoting a rumour to ridicule it may publicise it, so if you can, describe the rumour instead.
Headlines that treat an allegation as fact or that post an initial report of a viral claim without verifying it lend the rumour a second platform. Pachyphonus warned against grabbing on to supposed "hedge words" like "alleged", "according to police" and "sources said" just to lighten the blow - they reveal how much a statement is worth.
Conclusion
The Delhi story highlights a trend that defies borders or specific occurrences. Old footage, solicitous captions and unproven reports of damage are constantly propagated because they seem meaningful. The claims of well-planned foreign accounts are "unsubstantiated allegations", and the proven age of the returned content is "old archive footage". The real stickler is maintaining those categories. Before rushing to share the next scary footage, consider who's sharing it, when it was shot and who else has verified it.
Sources
- India Today: Anti-CEC protests, ISI-linked Pakistan accounts and false claims (report this piece is based on)
- The Week: Several Pakistani social media handles under scanner for fake content
- India TV: Delhi Police takes down 500 social media posts
- NewsMeter: Videos are old, unrelated to anti-CEC protests
- NewsMeter: Video of alleged attack on residence is from 2019
- Times of India (2020): How Delhi cops tried to rescue their colleagues
- Delhi Police statement on X, 12 October 2026
- Vosoughi, Roy and Aral (2018), The spread of true and false news online, Science
- Wardle and Derakhshan (2017), Information Disorder, Council of Europe
- International Fact-Checking Network, Code of Principles

Executive Summary
A video circulating on social media falsely claims to show West Bengal leader Suvendu Adhikari making inflammatory and communal remarks. In the viral clip, he is allegedly heard saying that if elected, his party would “work only for Hindus” and “eliminate the 27% Muslim population in West Bengal,” while also claiming that the BJP would remain in power for over 100 years. CyberPeace Research Wing research found that the viral video, which allegedly shows West Bengal leader Suvendu Adhikari making a statement about eliminating the Muslim population in West Bengal to ensure long-term BJP rule, is AI-generated and not real.
Claim
An X user (@Naira_Seth1) shared the clip alleging it reflects “majoritarian extremism disguised as politics,” suggesting that Suvendu Adhikari made communally charged and unconstitutional statements.

Fact Check
A detailed analysis of the video shows clear signs of manipulation. The person appearing in the latter part of the clip does not match Suvendu Adhikari.

Additionally, the video was tested using an AI detection tool (Hive Moderation), which indicated a 98.8% probability of AI-generated content, particularly around the 0:25 timestamp.

Further verification through reverse image search of key frames led to an original press byte of Suvendu Adhikari posted by PTI News on May 6. In that authentic video, he is seen speaking about taking strict legal action against criminals, with no mention of any religious community or communal statements.
https://x.com/PTI_News/status/2052020784152068114?s=20

Conclusion
The viral video allegedly showing Suvendu Adhikari making communal remarks is AI-generated and misleading. The original footage confirms that no such statements were made, and the clip has been digitally manipulated to spread misinformation.

About Global Commission on Internet Governance
The Global Commission on Internet Governance was established in January 2014 with the goal of formulating and advancing a strategic vision for Internet governance going forward. Independent research on Internet-related issues of international public policy is carried out and supported over the two-year initiative. An official commission report with particular policy recommendations for the future of Internet governance will be made available as a result of this initiative.
There are two goals for the Global Commission on Internet Governance. First, it will encourage a broad and inclusive public discussion on how Internet governance will develop globally. Second, through its comprehensive policy-oriented report and the subsequent marketing of this final report, the Global Commission on Internet Governance will present its findings to key stakeholders at major Internet governance events.
The Internet: exploring the world wide web and the deep web
The Internet can be thought of as a vast networking infrastructure, or network of networks. By linking millions of computers worldwide, it creates a network that allows any two computers, provided they are both online, to speak with one another.
The Hypertext Transfer Protocol is the only language spoken over the Internet and is used by the Web to transfer data. Email, which depends on File Transfer Protocol, Usenet newsgroups, Simple Mail Transfer Protocol, and instant messaging, is also used on the Internet—not the Web. Thus, even though it's a sizable chunk, the Web is only a part of the Internet [1]. In summary, the deep Web is the portion of the Internet that is not visible to the naked eye. It is stuff from the World Wide Web that isn't available on the main Web. Standard search engines cannot reach it. More than 500 times larger than the visible Web is this enormous subset of the Internet [1-2].
The Global Commission on Internet Governance will concentrate on four principal themes:
• Improving the legitimacy of government, including standards and methods for regulation;
• Promoting economic innovation and expansion, including the development of infrastructure, competition laws, and vital Internet resources;
• Safeguarding online human rights, including establishing the idea of technological neutrality for rights to privacy, human rights, and freedom of expression;
• Preventing systemic risk includes setting standards for state behaviour, cooperating with law enforcement to combat cybercrime, preventing its spread, fostering confidence, and addressing disarmament-related issues.
Dark Web
The part of the deep Web that has been purposefully concealed and is unreachable using conventional Web browsers is known as the "dark Web." Dark Web sites are a platform for Internet users who value their anonymity since they shield users from prying eyes and typically utilize encryption to thwart monitoring. The Tor network is a well-known source for content that may be discovered on the dark web. Only a unique Web browser known as the Tor browser is required to access the anonymous Tor network (Tor 2014). It was a technique for anonymous online communication that the US Naval Research Laboratory first introduced as The Onion Routing (Tor) project in 2002. Many of the functionality offered by Tor are also available on I2P, another network. On the other hand, I2P was intended to function as a network inside the Internet, with traffic contained within its boundaries. Better anonymous access to the open Internet is offered by Tor, while a more dependable and stable "network within the network" is provided by I2P [3].
Cybersecurity in the dark web
Cyber crime is not any different than crime in the real world — it is just executed in a new medium: “Virtual criminality’ is basically the same as the terrestrial crime with which we are familiar. To be sure, some of the manifestations are new. But a great deal of crime committed with or against computers differs only in terms of the medium. While the technology of implementation, and particularly its efficiency, may be without precedent, the crime is fundamentally familiar. It is less a question of something completely different than a recognizable crime committed in a completely different way [4].”
Dark web monitoring
The dark Web, in general, and the Tor network, in particular, offer a secure platform for cybercriminals to support a vast amount of illegal activities — from anonymous marketplaces to secure means of communication, to an untraceable and difficult to shut down infrastructure for deploying malware and botnets.
As such, it has become increasingly important for security agencies to track and monitor the activities in the dark Web, focusing today on Tor networks, but possibly extending to other technologies in the near future. Due to its intricate webbing and design, monitoring the dark Web will continue to pose significant challenges. Efforts to address it should be focused on the areas discussed below [5].
Hidden service directory of dark web
A domain database used by both Tor and I2P is based on a distributed system called a "distributed hash table," or DHT. In order for a DHT to function, its nodes must cooperate to store and manage a portion of the database, which takes the shape of a key-value store. Owing to the distributed character of the domain resolution process for hidden services, nodes inside the DHT can be positioned to track requests originating from a certain domain [6].
Conclusion
The deep Web, and especially dark Web networks like Tor (2004), offer bad actors a practical means of transacting in products anonymously and lawfully.
The absence of discernible activity in non-traditional dark web networks is not evidence of their nonexistence. As per the guiding philosophy of the dark web, the actions are actually harder to identify and monitor. Critical mass is one of the market's driving forces. It seems unlikely that operators on the black Web will require a great degree of stealth until the repercussions are severe enough, should they be caught. It is possible that certain websites might go down, have a short trading window, and then reappear, which would make it harder to look into them.
References
- Ciancaglini, Vincenzo, Marco Balduzzi, Max Goncharov and Robert McArdle. 2013. “Deepweb and Cybercrime: It’s Not All About TOR.” Trend Micro Research Paper. October.
- Coughlin, Con. 2014. “How Social Media Is Helping Islamic State to Spread Its Poison.” The Telegraph, November 5.
- Dahl, Julia. 2014. “Identity Theft Ensnares Millions while the Law Plays Catch Up.” CBS News, July 14.
- Dean, Matt. 2014. “Digital Currencies Fueling Crime on the Dark Side of the Internet.” Fox Business, December 18.
- Falconer, Joel. 2012. “A Journey into the Dark Corners of the Deep Web.” The Next Web, October 8.
- Gehl, Robert W. 2014. “Power/Freedom on the Dark Web: A Digital Ethnography of the Dark Web Social Network.” New Media & Society, October 15. http://nms.sagepub.com/content/early/2014/ 10/16/1461444814554900.full#ref-38.