A.I.
A.I.
A torrent of fake videos and images generated by artificial intelligence have overrun social networks during the first weeks of the war in Iran.
The videos — showing huge explosions that never happened, decimated city streets that were never attacked or troops protesting the war who do not exist — have added a chaotic and confusing layer to the conflict online.
The New York Times identified over 110 unique A.I.-generated images and videos from the past two weeks about the war in the Middle East. The fakes covered every aspect of the fighting: They falsely depicted screaming Israelis cowering as explosions ripped through Tel Aviv, Iranians mourning their dead and American military vessels bombarded with missiles and torpedoes.
Collectively, they were seen millions of times online through networks like X, TikTok and Facebook, and countless more times within private messaging apps popular in the region and around the world.
The Times identified the A.I. content by checking for both obvious signs — such as depictions of buildings that do not exist, garbled text and behaviors or movements that defy expectations — and for invisible watermarks embedded within the files. The posts were also checked with multiple A.I. detector tools and compared with reports from news organizations.
A sophisticated new wave of A.I. tools makes the fakes possible, enabling nearly anyone to create lifelike simulations of war that can deceive the naked eye for little to no cost. Similar content has spread in other conflicts, including the war between Ukraine and Russia. But this war has multiple fronts, and that has led to a proliferation of fake content since the United States and Israel first attacked Iran, according to experts.
“Even compared to when the Ukraine war broke out, things now are very different,” said Marc Owen Jones, an associate professor of media analytics at Northwestern University in Qatar. “We’re probably seeing far more A.I.-related content now than we ever have before.”
Overall, the A.I. fakes included …
37 fake images and videos falsely depicting active war
5 fake images and videos falsely depicting war preparation
8 fake images and videos falsely depicting destruction
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
5 fake images and videos falsely depicting crying soldiers
43 memes and overt uses of A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
13 other fake images and videos
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
A.I.
The content has become a potent informational weapon for Tehran as it seeks to shake the public’s tolerance for war by depicting scenes of devastation and destruction across the region. The majority of A.I. videos about the war push pro-Iranian views, often to falsely demonstrate its military superiority and sophistication, according to a study of online activity by Cyabra, a social media intelligence company.
“The use of A.I. images of places in the Gulf — being burnt or damaged — becomes more important in Iran’s playbook,” Mr. Jones said, “because it allows them to give a sense that this war is more destructive and maybe more costly for America’s allies than it might actually be.”
In one of the most circulated fake videos found online, a shaky handheld scene seemingly shot from an apartment balcony in Tel Aviv shows the skyline pounded with missiles as an Israeli flag sits in the foreground. The video was viewed millions of times across platforms and was picked up by social media influencers and fringe news websites, according to a review of social media activity by The Times.
The Israeli flag in the foreground was one telltale sign that the video was A.I.-generated, experts said. To generate such videos, creators who use A.I. tools will typically write simple text instructions describing, for example, a shaky handheld video of a missile strike on Israel. The A.I. tools will then often include an Israeli flag or the Star of David to fulfill such a request. Several other A.I. videos included the flag.
There is ample genuine footage of the war being shared online, too, with cellphones and social platforms giving a real-time view of the conflict. Many of those images and videos are more subdued than the scenes made by A.I. tools.
Real footage of missile strikes was often shot from far away, typically at night, with missiles visible as little more than bright lights in the distance. Explosions in real videos are more often shown as plumes of smoke, not as fireballs, with bystanders rushing to film the scene only after the munitions meet their target.
Some A.I. videos and images, by contrast, have falsely depicted war like an over-the-top Hollywood action movie, with enormous explosions resulting in mushroom clouds, sonic booms that ripple across unnamed cities and supposed hypersonic missiles that leave glowing streaks in the sky. Real footage is sometimes enhanced by A.I. tools to make explosions appear larger and more devastating, further blurring the line between what is real and fake.
The A.I. footage has essentially created an alternate reality more suited to social media, experts said, where the exaggerated footage is more likely to find an audience.
In one instance, the A.I. fakes played an outsize role in the debate online and between governments over the fate of the U.S.S. Abraham Lincoln, an aircraft carrier deployed to the region. Iran’s Islamic Revolutionary Guards Navy initially suggested on March 1 that they had successfully attacked the ship, possibly sinking it. That led to a deluge of A.I.-generated fakes depicting the ship or those like it on fire. Iranian users celebrated the footage online as evidence that their country’s counteroffensive was rattling the U.S.-Israeli alliance.
The United States later said that the attack was unsuccessful and that the ship was unharmed.
A.I.
Dozens of other A.I. images and videos made no effort to hide that they were fake, acting instead as a new form of digital propaganda that brought to life the political arguments typically made by governments or their propaganda arms. Those included flattering depictions of world leaders as powerful men, or dehumanizing depictions of opposition leaders.
One collection of clearly fictional videos offered a view of the Shajarah Tayyebeh elementary school, which was destroyed by the United States in an apparent errant missile strike on Feb. 28, according to a preliminary inquiry. At least 175 people were killed, most of them children, according to Iranian officials.
The A.I.-generated videos unfolded like short films, showing school girls playing outside before an American fighter jet launches missiles.
Social media companies have done little to combat the scourge of A.I. videos that overwhelmed their platforms last year after OpenAI released Sora, a video-generating app that allowed anyone to create realistic fakes through a simple app. (The New York Times sued OpenAI and Microsoft in 2023, accusing them of copyright infringement of news content related to A.I. systems. The two companies have denied those claims.)
Though videos generated by many A.I. tools can include both visible and invisible watermarks labeling them as fake, those are easy to remove or obscure. Only a few of the videos identified by The Times contained such watermarks.
Elon Musk’s X, which has taken a broadly permissive approach to allowing misinformation on its platform, announced last week that it would suspend accounts from receiving revenue from the platform for 90 days if they posted A.I.-generated content of “armed conflict” without labeling it as such, in a bid to stop users from profiting off the falsehoods.
But many of the Iranian-linked accounts identified by Cyabra appeared far more focused on spreading its messages than making money.
“This is a natural front for Iran to try and exploit and it feels like this is one of the reasons it is so voluminous,” said Valerie Wirtschafter, a fellow at the Brookings Institution studying foreign policy and A.I. “It’s actually a tool of war.”
