A family Hanukkah gathering at Bondi Beach became the scene of a terror attack that left at least 16 dead and dozens more injured. In the grainy, frantic videos that raced ahead of any verified reporting, one figure stands out: Ahmed al-Ahmed-an unarmed bystander who ran toward the gunfire, wrestled a long gun away from an attacker, and was shot and injured in the process.
In a healthier information environment, that should have been the crux of the story: a Syrian-Australian Muslim, a father, a shopkeeper, doing what conscience exceptionally demands and what most of us only applaud after the fact. But grief, panic, and politics rarely leave heroism unsoiled. Within hours, Ahmed’s identity began to be rewritten in real time-not by historians, but by a swarm of websites, accounts, and automated amplification.
First, he was transformed into an Anglo-Australian “Edward Crabtree,” a Bondi local-complete with an invented biography and fabricated details-published by a website that appeared the very day of the attack. Then, a second draft, feeding the appetites of a different faction, claimed he was not Muslim at all but a Lebanese Maronite Christian- a falsehood repeated widely enough that major outlets had to note it and correct it.
This isn’t random. It is targeted identity politics with sinister motivations and murderous goals.
When the hero is a Syrian Muslim who saved Jews at a Jewish event, he becomes inconvenient to individuals and ideologies thirsty for clear, absolute categories. So the moral narrative gets edited until it fits the script. A Muslim hero becomes a “real Aussie,” and if that doesn’t satisfy, he becomes a Christian-anything, apparently, except what he is.
Why did the lie travel faster than the truth? Part of the answer is ominous: we have built an ecosystem in which mainstream sources are distrusted-sometimes justifiably, too often with irrational certainty. In the first hours after a catastrophe, credible institutions move cautiously, while feeds reward certainty. Even “helpful” tools collapse under pressure: AAP documented how Grok, X’s chatbot, misrepresented a viral clip of Ahmed’s disarming of a gunman with an absurdly wrong explanation, adding yet another layer of noise.
But ideology alone does not explain the scale and coordination of modern disinformation. Democracies are vulnerable precisely because they are open-open networks, open platforms, open argument. That openness is a strategic target. European and U.S. security reporting has repeatedly warned that foreign actors use disinformation-now increasingly aided by generative AI-to undermine trust, polarize populations, and weaken democratic resilience.
The aim is not to make people believe one lie; it is to make them believe nothing-to make truth feel like a partisan costume, and to train citizens to fear everyone except their own kith and kin.
This is where deepfakes shift from “media weirdness” to civilizational hazard. The World Economic Forum has been blunt: misinformation and disinformation are among the most severe near-term global risks, and synthetic content accelerates the problem. European institutions are equally blunt about how generative AI can supercharge information manipulation.
I understood this more viscerally after reading antisemitic comments under Italian news articles-antisemitic comments so polished, so psychologically literate, so engineered to seduce the reader into “reasonable” hatred, that it was hard to believe they were merely the hobby of bored bigots. Perhaps they were. Or perhaps the bigots are being coached, nudged, scaled.
Now imagine what comes next: exceptionally realistic clips of imaginary Hasidic rabbis at a Mea Shearim tish cursing the Pope; imaginary cheder children in Brooklyn joking about Southerners as dirty cockroaches; or an imaginary French or British imam inciting the crucifixion of baptized children-videos designed not to persuade the already convinced, but to tip the hesitant into disgust.
Whether the production sits in Moscow, Beijing, or elsewhere, distribution can be outsourced to the same attention economy that turned Ahmed into Edward Crabtree. And the targets will not be only Jews: immigrants, whites, blacks-any group that can be made to look monstrous on a loop. The escalation from imaginary scenes to real stabbings, real lynchings, and riots and pogroms that sink entire nations into civil war is rapid. The 20th-century sectarianism and strife of India, Lebanon, Yugoslavia, and Rwanda chronicle the tragic pattern.
The Gaza war showed something even darker: the nations are not just horrified when Jews do something that looks or can be made to look bad; they seem to relish their self-righteous rage. Deepfakes are gasoline in a room already heavy with fumes.
What, then, is the antidote? Censorship and regulation are unworkable. I propose inoculation through exposure: flooding the internet with deepfakes so that future deepfakes lose their power-especially via poorly produced deepfakes portraying rabbis, priests, imams, and others saying absurd things that are toxic and ludicrous rather than toxic and hateful.
Without this vaccine that discredits future deepfakes of this sort, it is only a matter of time until the Russians, Chinese, or Qataris provide Tucker Carlson and Candace Owens with deepfakes that cause more horrible damage than even the Protocols of the Elders of Zion.
Israel’s decision-makers need to act. Urgently. Otherwise, AI may become yet another gift of Jewish genius that blesses everyone, save Am Yisrael.
Rafael Castro is a political analyst who studied at Yale and Hebrew University. An Italian-Colombian Noahide by choice, Rafael can be reached at rafaelcastro78@gmail.com