Israel-Iran conflict unleashes wave of AI disinformation

A wave of disinformation has been triggered online since Israel started the strikes on Iran last week, with dozens of messages examined by the BBC verification aimed at amplifying the efficiency of Tehran’s response.

Our analysis has found a number of videos – created using artificial intelligence – boasting of military capacities of Iran, alongside false clips showing the consequences of strikes on Israeli targets. The three fake most visualized videos BBC Verify found collectively amassed more than 100 million views on several platforms.

Pro-Israeli accounts have also shared online disinformation, mainly in recirculation of old clips of protests and gatherings in Iran, wrongly affirming that they show increasing dissent against the government and the support of Iranians for the Israeli military campaign.

Israel launched strikes in Iran on June 13, leading to several cycles of Iranian missiles and drone attacks against Israel.

An organization that analyzes open source imaging describes the online disinformation volume as “astonishing” and accused some “engagement farmers” of seeking to take advantage of the conflict by sharing a deceptive content designed to attract attention online.

“We see everything, from the unrelated images of Pakistan to the recycled videos of the October 2024 strikes – some of which have raised more than 20 million views – as well as the game clips and the content generated by AI -A -TRANSISM as real events,” wrote GeoconfirMed, the online verification group, wrote on X.

Some accounts have become “super-repairing” disinformation, rewarded with significant growth in their number of followers. A pro -Iranian account without obvious links with the authorities of Tehran – Iran’s daily military – saw its subscribers on X go from just over 700,000 on June 13 to 1.4 m in June, an increase of 85% in less than a week.

These are many obscure accounts that have recently appeared in people’s flows. All have blue ticks, are prolific in mailbox and have repeatedly displayed disinformation. Because some use apparently official names, some people assumed that they are authentic accounts, but it is not clear that really manages profiles.

The disinformation torrent has marked “the first time that we have seen a generative AI be used on a large scale during a conflict,” BBC Verific Emmanuelle Saliba, director of the analysts’s group, told BBC.

The accounts examined by the BBC check the frequently shared images generated by AI which seem to seek to exaggerate the success of Iran’s response to Israel strikes. An image, which has 27 million views, represented dozens of missiles falling on the city of Tel Aviv.

Another video supposed to show a missile strike on a building in the Israeli city late at night. Ms. Saliba said clips often represent night attacks, which makes them particularly difficult to check.

AI counterfeits have also focused on the demands for destruction of Israeli F-35 fighter planes, a cutting-edge art of art capable of hitting targets on the ground and in the air. If the clip barrage was the real Iran would have destroyed 15% of the fighters’ fleet of Israel, Lisa Kaplan, CEO of the Alethea analyst group, told BBC Verify. We have not yet authenticated F-35 images in the process of taking.

A largely shared position claimed to show a damaged jet after being shot in the Iranian desert. However, the signs of manipulation of the AI ​​were obvious: civilians around the jet were the same size as the vehicles nearby, and the sand showed no sign of impact.

Another video with 21.1 million views on Tiktok has said that an Israeli F-35 is shot down by air defenses, but the images came from a flight simulator video game. Tiktok deleted the images after being approached by BBC Verify.

Ms. Kaplan said that part of the accent on the F-35 was fired by a network of accounts that Alex has already linked to Russian influence operations.

She noted that the operations of Russian influence recently passed the course of the tension to undermine support for war in Ukraine to doubts about the capacity of Western weapons – in particular American -.

“Russia does not really have an answer to the F-35. So what can he do? He can seek to undermine his support in certain countries,” said Kaplan.

Disinformation is also distributed by well-known accounts that have already weighed on the War of Israel-Gaza and other conflicts.

Their motivations vary, but the experts have said that some perhaps try to monetize the conflict, certain major social media platforms offering account payments to reach a large number of views.

On the other hand, pro-Israeli posts have largely focused on the suggestions that the Iranian government is confronted

Among them is a video generated with widely shared that wrongly claims that the Iranians sing “we love Israel” in the streets of Tehran.

However, in recent days – and as speculation on American strikes on Iranian nuclear sites have been developing – some accounts have started to publish images generated by AI of B -2 bombers on Tehran. The B-2 has drawn special attention since the start of Israel’s strikes on Iran, because it is the only plane capable of effectively making a Attack on Iranian underground nuclear sites.

The official sources in Iran and Israel have shared some of the false images. The state media in Tehran shared false strikes of strikes and an image generated by the AI ​​of a slaughtered F-35 jet, while a post shared by the defense forces Israel (IDF) received a community note on X for having used old images not linked to missile dams.

A large part of the disinformation examined by BBC Verify was shared on X, users frequently turning to The AI ​​chatbot of the platform – Grok – To establish the veracity of messages.

However, in some cases, Grok insisted that AI videos were real. Such a video has shown an endless flow of trucks carrying ballistic missiles emerging from a complex on the mountainside. The indicative signs of AI content included rocks in the moving video of their own will, said Saliba.

But in response to X users, Grok has repeatedly insisted that the video was real and cited media reports, including Newsweek and Reuters. “Check the news of confidence for more clarity,” concluded the chatbot in several messages.

X did not respond to a request from the BBC Check the comments of the actions of the chatbot.

Many videos have also appeared on Tiktok and Instagram. In a declaration at BBC Verify, Tiktok said that it proactively applies the directives of the community “which prohibit inactive, deceptive or false content” and that it operates with independent fact verifiers to “check the deceptive content”.

The owner of Instagram Meta did not respond to a request for comments.

Although the motivations of those who create online counterfeits vary, many are shared by ordinary social media users.

Matthew Facciani, a researcher at the University of Notre Dame, suggested that disinformation can spread more quickly online when people are faced with binary choices, such as those raised by conflicts and politics.

“This testifies to the broader social and psychological question of people who wish to re-party things if it aligns with their political identity, and also in general, the more sensationalist emotional content will propagate faster online.”