Imagine waking up to a message in the family WhatsApp group. Your cousin has shared a photo of burning buildings in Tel Aviv, with a caption saying the government is hiding the true scale of destruction after an Iranian strike. Your stomach drops. You feel anger, fear, helplessness. You share it too. But the photo was not taken by a journalist or a satellite. It was generated by an artificial intelligence tool, in flawless Hebrew, in a matter of seconds, by a network whose entire purpose is to make Israelis feel exactly that way. This is the new front line of Israel’s security, and it does not look like a missile or a drone. Generative artificial intelligence has removed the last technical barriers for Israel’s adversaries. Large language models produce targeted content in perfect Hebrew within seconds, at a price close to zero. The technology is not secret; anyone with an internet connection and a laptop can use it to create convincing images, videos, and articles. During the June 2025 strikes on Iran, a network called Isnad, founded by an exiled Egyptian activist, spread doctored images of burning buildings that exaggerated the destruction in Israel, aiming to break public morale. Iranian networks run most of the campaigns tracked inside Israel, with some Russian activity behind them. The reach is real. Nearly one in five identified disinformation campaigns succeeded in planting material that Israeli media then reported as genuine news. FakeReporter’s researchers describe the mechanism plainly: interference networks radicalize people, draw out reactions that might otherwise stay private, then amplify them. The World Economic Forum ranked disinformation as the world’s top short-term global risk of 2025, and Israel is living that ranking. But most Israelis do not realize they are living inside a psychological operation. They think they are just scrolling, sharing, and reacting like normal people. That is exactly the point. The attack is not aimed at military bases or power grids. It is aimed at the mind, at the way people interpret events, at the trust that holds a society together. A single fabricated image can travel from a Telegram channel to a national news broadcast in hours, and by the time anyone checks it, the emotional damage is already done.
The strange part is that Israel, a country that prides itself on being a startup nation, has no real answer to any of this. The state comptroller reported in June that the country is completely unprepared for foreign influence operations. There is no national policy and no authority responsible for the threat. It is as if a country that built some of the most advanced cyber defenses in the world forgot to lock the door to the living room. Siman-Tov, a senior researcher at the Institute for National Security Studies in Tel Aviv, calls this a conceptual failure: the state still protects infrastructure when it needs to start protecting the cognitive space where citizens actually form their opinions. Think about what that means. The government knows how to protect electricity grids, water supplies, and military command centers. But it has not built an institution to protect the mental space where a mother decides whether her country is safe, where a soldier interprets what he sees on his phone, where a voter decides whether an election is legitimate. There is movement behind the silence. The Shin Bet reportedly now ranks foreign interference as a top-level threat and has sent staff to countries facing similar risks to study their methods. But an intelligence service alone cannot guard what happens in a family WhatsApp group, and that is where most Israelis meet the lie first. The threat does not arrive in uniform. It arrives as a forwarded message from someone you love. It arrives as a meme that makes you angry. It arrives as a headline that confirms your worst fears. By the time an intelligence agency notices it, it has already passed through thousands of hands and settled into the way people talk and think. This is not a technical problem with a technical solution. It is a social and institutional problem, and Israel has not even admitted it is one. The silence from the top is not a sign of calm; it is a sign of absence. A country that can intercept a rocket in mid-air cannot intercept a lie that has already been shared by a thousand people.
The uncomfortable truth is that Israel is not alone in this exposure, and some of its friends have already started building defenses. Taiwan has spent years building rapid-response teams and civic tech tools after absorbing waves of pressure campaigns from Beijing, and its playbook is studied in Washington and Brussels. Sweden weathered a sustained campaign before its NATO accession and rebuilt parts of its psychological defense around civilian institutions rather than intelligence services. What do these countries have in common? They treat the problem as a permanent feature of modern life, not a temporary crisis. Their defense is slow, boring, and institutional. It is a standing capability with a budget line and a boss. It is not a heroic rescue operation that swings into action when a crisis hits. It is a permanent part of the state, like a fire department or a public health system, that works quietly all the time. It has a minister or an agency that owns the mission, and it does not disappear when the news cycle moves on. Taiwan trains ordinary citizens to spot influence operations and gives them tools to verify information before sharing it. Sweden runs public education campaigns that teach people how propaganda works, not just what the latest fake looks like. These are not glamorous projects. They do not generate dramatic headlines. But they build a kind of immunity in the population, and that immunity is exactly what Israel lacks. The country that invented the Iron Dome has no equivalent for the cognitive layer. It has no national authority for psychological defense, no rapid-response team for viral lies, no public education system for digital literacy. It has not even agreed on who should be responsible for building such a system. It is the one front in this war where Israel never built anything. The river keeps rising whether or not anyone agrees it is a flood. The countries that take the threat seriously are not the ones that shout the loudest. They are the ones that quietly built the boring institutions that make a population harder to manipulate. Israel, which has never lacked for courage or creativity, has somehow missed this lesson. The Israeli instinct is to invent a brilliant solution and move on; this threat requires the opposite.
The economics of this fight are even worse than the politics. An influence operation can be launched from a laptop anywhere for the price of an afternoon. The material it produces never expires. A doctored clip forged in June resurfaces in September with a new caption, and the fact-check written the first time does not travel with it. It starts fresh, like a virus that mutates just enough to fool the immune system again. Defense is the opposite. Every fake has to be found, documented, rebutted, and then found again when it comes back. It is an endless game of whack-a-mole, and the attacker only needs to win once while the defender has to win every time. Small watchdog outfits like FakeReporter are doing work that properly belongs to a standing national institution, on budgets that would not cover a single missile interceptor. They are staffed by people who believe in what they are doing, but they cannot scale to meet a threat that grows cheaper and faster with every improvement in AI. Meanwhile, the platforms where most of this content spreads have their own incentives, and they are not the same as a country’s security interests. A viral lie is good for engagement. A shocking image is good for ad revenue. The platforms are not neutral; they are the delivery system for the attack. The algorithms do not care whether the content is true; they care whether it stops people from scrolling. So the burden falls on the people who see the lie, and on a handful of underfunded organizations trying to catch it before it does too much damage. Even when a debunking is published, it reaches a fraction of the audience that saw the original lie, and it lacks the emotional punch that made the lie spread. That is not a strategy. It is a desperate improvisation. And the asymmetry is only going to get worse. As AI improves, the cost of creating a convincing lie will fall even further, while the cost of verifying every piece of information will stay high. The defender is running a marathon on a treadmill that gets faster every day. There is no finish line, no moment when the defense can declare victory and go home.
The honest answer is that the most effective defenses so far are the smallest ones. FakeReporter and similar civil society monitors find the fake networks faster than any ministry does. Newsrooms that check before amplifying have started to close the laundering route that one in five campaigns depended on. And the individual habit holds more power than it gets credit for: pause before sharing something engineered to provoke. Readers who want to check what they are forwarding can use verified reference material instead of the feed that served them the fake. But there is a next stage that is far more dangerous, and Siman-Tov warned of it. He calls it data contamination. Hostile actors are beginning to pollute the information that AI models are trained on, so the falsehood comes pre-baked into the tools people trust for answers. The goal of all of it, he noted, is not necessarily to change how people vote. It is enough to plant the idea that the election is rigged. That warning deserves more attention than it has received, because it moves the battlefield somewhere no fact-checker patrols. When a hostile network seeds false statistics and forged documents into the open material that language models learn from, the falsehood stops arriving as a suspicious stranger and starts arriving as a memory. It feels like something you have always known, because the model presents it with the same authority as a date in history. A student asks a chatbot about an Israeli election incident and receives, in confident prose, an invented event that traces back to Tehran’s fabrication mills. No alert sounds. The lie has been laundered not through a newsroom but through the archive itself, and correcting it means chasing copies of a training corpus nobody controls. Israel’s research community has begun cataloging these poisoned sources, but the effort is scattered across universities and volunteer groups with no mandate, no budget, and no authority over the platforms that serve the answers to billions of people. This is the hardest part of the problem, because it is invisible. People do not know what they do not know. A model that has been poisoned does not announce its bias; it simply presents the falsehood as fact, with the same calm confidence it uses for everything else.
There is a quiet precedent for fixing this. Israel once treated cybersecurity as a niche concern for banks and the military, until sustained attacks forced the creation of a national authority with real powers. The same recognition now has to arrive for the cognitive layer. The cost of delay compounds quietly. Every month without verification standards, every newsroom that publishes first and checks later, and every model trained on unexamined material makes the next campaign cheaper for the adversary and dearer for the defense. This is not a problem that can be solved by buying more weapons or building another wall. It requires a national institution with a clear mandate, a permanent budget, and the authority to coordinate across ministries, platforms, universities, and civil society. It requires public education that treats digital literacy as a basic skill, like reading and writing. It requires news organizations to treat verification as a sacred duty, not a cost center. And it requires ordinary people to understand that they are not just consumers of information but targets of a campaign. The river keeps rising whether or not anyone agrees it is a flood. Israel built its defenses for an ocean of missiles. It is now facing a river that never stops flowing, in its own language. The seawall was designed for a different sea. The cost of inaction is not measured in missiles or budgets; it is measured in the slow erosion of a shared reality. When people can no longer agree on what is true, democracy becomes impossible. The question is not whether the next campaign will come. It is whether Israel will finally build the defense it has never had, before the river rises high enough to wash away the trust that holds the country together. That trust is the real target, and it is the one thing no missile defense can protect. The technology will keep improving. The adversaries will keep adapting. The only question is whether the people who run the country will wake up to the fact that the war is already inside the walls, speaking their language, wearing their faces, and living in their phones. The next campaign is already being designed, in some language model, by someone who knows exactly how to make Israelis doubt their own eyes.