adam zivo: iran’s desperate supporters use deepfakes to fabricate victory
since they can’t win in reality, they're using ai to hallucinate fantasies of israel's destruction
4 minute read
a man takes a selfie while sitting on a beach chair while taking shelter in a parking garage during a missile alert from iran, in tel aviv, on june 20.
bernat armangue/ap photo
while iran has managed to land some missile strikes upon israel, these attacks have neither crippled the country nor paralyzed israeli life. tehran and its supporters do not like this, though, so they are using shameless lies — including ai-generated content — to fabricate an alternative digital reality where israeli cities lie in ruins.
i am currently reporting from israel on behalf of the news forum (a canadian news channel), and can confidently say that the damage here has been relatively limited, thanks to the iron dome defence system, and that the israelis are, at least for now, clearly winning this war.
that doesn’t mean that israel has been immune to harm: several buildings have been damaged and destroyed, including a major hospital in the southern city of beersheva, and around 24 israelis have been killed, with thousands more injured and several thousand displaced or rendered homeless.
however tragic these losses may be, though, they are a far cry from the cataclysm tehran had originally hoped to inflict. the ayatollah wanted to see tel aviv reduced to rubble through a retaliatory strike of 1,000 ballistic missiles — but the israel defence forces (idf) decapitated his air force and destroyed over a third of his missile launchers. now, israeli aircraft dominate his skies and only a fraction of iran’s stockpiles have been utilized, with new strikes apparently diminishing in scope.
advertisement
to put things into perspective another way: a former senior israeli intelligence officer, miri eisin, told an international media outlet earlier last week that israeli prime minister benjamin netanyahu’s government had anticipated that iran would kill around 5,000 civilians through its initial strike. so far, actual fatalities have amounted to less than one per cent of that figure.
tehran and its supporters are losing, but they cannot accept this, so they are flooding the internet with posts that falsely claim that the damage in israel is far more extensive than the media is reporting.
“don’t be fooled, israel is in worse shape (than) people think. about 1/3 of tel aviv has been damaged or destroyed,” posted douglas macgregor, a once-respected military expert, to x on friday. despite being obviously unhinged, his fan-fiction garnered millions of views and tens of thousands of likes.
oftentimes, this kind of disinformation uses mislabelled photos and videos from other disasters or conflict zones. missiles and explosions from ukraine and gaza are falsely presented as having occurred in israel, for example. while this is not a new tactic — during the war against hamas, images of old horrors in syria were often misrepresented as gazan realities — the increasing prevalence of ai-generated content has been a game-changer.
advertisement
some of these ai posts haven’t been exactly subtle. for example, a pro-china x account named america-china watcher posted an ai video last thursday that reportedly showed a “before and after” of iran’s alleged destruction of tel aviv. the explosions were cheesy and obviously fake, and the video’s mise-en-scène didn’t make sense. nonetheless, the post went viral and was viewed over 200,000 times.
on june 17, a popular far-right x account called white ghost posted an image of missiles raining down like hellfire on a nighttime city. “this is not ai. this is tel aviv,” claimed the caption — but the bbc quickly verified that the image was, in fact, ai-generated and had been stolen from a digital creator’s facebook post. still, the post received over 230,000 likes and was viewed over 27-million times.
even iran’s legacy media has embraced ai forgeries: on june 15, the tehran times posted a video reportedly showing an iranian missile hitting a building in tel aviv. the video was quickly debunked, as it contained a watermark indicating that it had been produced by google’s ai video-generating system. yet, the post was not taken down and has received almost half a million views so far.
this is just a minuscule sample of the propaganda that is being disseminated by iran’s allies. if you can’t win in reality, you can just hallucinate a victory, apparently.
advertisement
while some of this content is being shared and promoted organically, it is also being boosted by iran’s professional disinformation agents. last year, radware, an israeli cybersecurity firm, documented how tehran is using ai to enhance its networks of social media bots (e.g., creating ai-generated personas and deepfakes) and influence public opinion. last wednesday, the firm released an update indicating that these networks have been deployed to support pro-iranian narratives in the current war.
several days ago, i posted a video of the beaches of tel aviv that debunked the “israel is destroyed” narrative by showcasing people resiliently socializing, exercising and playing volleyball. it quickly went viral, amassing millions of views, leading to a deluge of commentators claiming it was fake, outdated or ai-generated.
some of these people seemed to be bad actors and claimed, impossibly, that they had seen this footage months or years before. others seemed earnestly suspicious, and had simply lost the capacity to trust online content.
but something strange happened: many commentators asked ai chatbots to verify whether the video was authentic. when these chatbots, being unequipped for this task, gave inconsistent, conjecture-laden answers, this was taken as evidence of forgery. so not only is ai flooding the internet with deepfakes, it is being used as a shoddy fact-checker, too, and, in this way, has found yet another means to erode reality.
advertisement
national post
share story
