Daniel sent us this one — and it's a good one. In late May, Israel Hayom dropped a bombshell report revealing the existence of a secret Mossad branch dedicated entirely to influence operations against Iran. This unit was apparently stood up in 2021 under Mossad director David Barnea, and until now nobody outside the classified world even knew it existed. The branch was built to map Iranian society from the inside — public moods, media trends, generational fault lines — and then weaponize those vulnerabilities. The toolkit includes media leaks, coordinated social-media campaigns, and even AI-generated personas and influencers. The doctrine, as described, was that influence operations combined with kinetic strikes could help bring down the regime. And the posture, quoted in the piece, was "we are not done with Iran, we are just getting started." There's a lot here — and it pulls back the curtain on a side of Mossad tradecraft that gets way less attention than the explosions and assassinations.
The part that grabbed me immediately was the concrete case they disclosed — the takedown of Rostam Ghasemi. Ghasemi was a former oil minister, former commander in the IRGC, a heavyweight in the system. And according to Israel Hayom, the Mossad had been sitting on an embarrassing photograph of him since around 2011 — Ghasemi with a woman not wearing a hijab. They held it for over a decade, then quietly released it during the nationwide unrest in 2022. Ghasemi resigned within days. That's not a car bomb, that's not a motorcycle hit — it's a photograph released at the right moment, and it took out a regime figure more cleanly than any kinetic operation could have.
The slow knife. Eleven years in a drawer, then someone hits send at exactly the right week — and a career ends. There's something almost surgical about it that the explosions can't match.
It tells you something about how this branch actually operates day to day. It's not Hollywood spycraft — it's archivists, analysts, and patience. Someone catalogued that photograph in 2011. Someone maintained the file. Someone watched Iranian politics closely enough to know when Ghasemi was vulnerable, when releasing it would have maximum effect. That's not glamorous work. That's librarianship with geopolitical consequences.
Walk me through what this branch actually is, structurally. Because "influence operations" can mean anything from tweeting memes to running a fake news network with a hundred employees.
So the Israel Hayom piece describes a unit built to do three things. One: map Iranian society from the inside — public sentiment, media consumption patterns, what different demographics are angry about, where the pressure points are. Two: identify and cultivate vulnerabilities — individual compromising material like the Ghasemi photograph, but also structural weaknesses in how the regime communicates with its own population. Three: weaponize those vulnerabilities through coordinated campaigns — media leaks, social-media amplification, and now AI-generated content and personas. The branch reportedly recruited Persian-speaking operatives, data scientists, and psychological warfare specialists. It's essentially a fusion cell that combines open-source intelligence, human intelligence, and digital operations under one roof.
It's not a classic HUMINT branch, not a sabotage branch, not a targeted-killing branch. It's a fourth thing.
It's a fourth thing, and that's what makes the disclosure significant. We've known for years about Mossad's kinetic operations in Iran — the Natanz sabotage, the assassination of Mohsen Fakhrizadeh, the theft of the nuclear archive in 2018. Those are dramatic, they make headlines, they fit the action-movie template. But this branch is doing something different. It's trying to accelerate regime decay from within by manipulating the information environment. The goal isn't to destroy a centrifuge — it's to make Iranians believe the regime is weaker than it appears, to amplify internal dissent, to create the perception of inevitable collapse.
Which loops back to something I've always found interesting about psychological warfare — the target isn't the centrifuges, it's the belief that the centrifuges matter. If you can convince enough people the system is doomed, they start acting like it's doomed, and then it is.
That's the theory. And the Ghasemi case is a perfect small-scale demonstration. The photograph itself — a former IRGC commander with a woman not wearing a hijab — that's not illegal in some technical sense, but in the moral economy of the Islamic Republic it's devastating. Ghasemi built his career on revolutionary credentials, on piety, on adherence to the system's values. And then suddenly there's photographic evidence that he's a hypocrite — at exactly the moment when millions of Iranians are in the streets protesting the morality police. The timing is the weapon as much as the image.
Of course it is. The photograph in 2011 would have been a minor scandal — embarrassing, but survivable. The photograph in the middle of the Mahsa Amini protests is a career-ending event.
That's the operational discipline here. Most intelligence agencies, if they get compromising material on a foreign official, their instinct is to use it immediately — leverage them, turn them into an asset, extract something. The Mossad sat on this for eleven years because they understood the photograph would be worth more as a public weapon in the right context than as private blackmail. That's a long-term strategic mindset that most agencies struggle to maintain.
It's the intelligence equivalent of aging whiskey. You don't crack the barrel because you're thirsty.
I'm going to use that. But yes — and it raises the question of what else they're sitting on. If they had Ghasemi's photograph since 2011, what do they have on current officials? What's in the archive that hasn't been deployed yet?
That's actually part of the psychological effect of revealing the branch's existence at all, isn't it? You tell the world you have a unit dedicated to this, you disclose one case study, and suddenly every Iranian official is wondering whether there's a photograph of them in a drawer in Tel Aviv.
That's the skeptical caveat we need to address here. This disclosure came through Israel Hayom — a newspaper owned by Sheldon Adelson's family, closely aligned with Prime Minister Netanyahu, often used as a conduit for messaging the government wants out but doesn't want to officially own. The timing is May 2026 — right when Israeli strikes in Lebanon are escalating, the Strait of Hormuz is in the news, and US-Iran talks are teetering. Revealing that you have a secret influence-operations branch is itself an influence operation.
The revelation is the operation. You're not just telling the public — you're telling Tehran, "We have archives you don't know about, operatives you haven't found, and patience you can't match." It's a capability demonstration without firing a shot.
Capability demonstrations are a core influence-operations tactic. You don't have to prove you can do everything — you just have to make the target believe you can. The Israel Hayom piece is careful to include details that signal sophistication without revealing anything operationally useful. AI-generated personas. Coordinated social-media campaigns. Persian-speaking operatives. Those are signals to different audiences — to the Iranian public, "we're in your information space," to the regime, "we've penetrated your society," to allies and adversaries, "our capabilities are more advanced than you thought.
Let's talk about the AI-generated personas piece, because that's where this gets genuinely novel. We've covered AI in plenty of contexts on this show, but using language models to create fake Iranian activists, fake journalists, fake dissidents — that's a different order of problem.
The Israel Hayom report mentions it almost in passing, but it's potentially the most significant capability described. If you can generate dozens or hundreds of convincing Persian-language personas — each with a consistent posting history, a believable biography, a network of mutual followers — you can shape online discourse in ways that traditional bot farms never could. Traditional bots are detectable because they're repetitive, they post at inhuman hours, their language is stilted. A well-tuned language model generating unique, idiomatic Persian text for each persona, posting at realistic intervals, engaging with real users — that's much harder to spot.
The Turing test, but the stakes are regime change.
The ethics get very uncomfortable very fast. If the Mossad is creating fake Iranian activists who are calling for protests, and real Iranians join those protests and get arrested or killed — what's the moral calculus there? The AI persona didn't exist, but the real person who followed its call to action did.
That's the part where I pull back and ask whether we're describing a legitimate intelligence operation or something closer to entrapment. If you're manufacturing dissent that looks organic, and real people pay the price for acting on it, you're not amplifying genuine grievances — you're creating a false signal that gets real people hurt.
There's a counterargument that the grievances are already real — the regime's brutality, the economic collapse, the suppression of women's rights. The influence operation isn't fabricating the anger, it's giving it coordination and amplification. But I'm not sure that fully resolves the ethical problem. There's a difference between saying "the regime is brutal, here's how people are resisting" and creating a fake person who says "meet at Azadi Square at 6 PM, we're all going.
The fake person didn't risk anything. The real person who showed up did. And that asymmetry is where the ethical floor drops out.
It's the same problem that makes state-run disinformation so corrosive in any context. Even if the cause is just — and I'd argue that weakening the Iranian regime is a just cause — the method erodes something fundamental. If Iranians can't trust that the voices they're hearing online are real, if every dissident account might be a Mossad AI or an IRGC counter-operation, then the public sphere becomes unreadable. Everyone retreats into suspicion. That's not a foundation for democratic revival — it's a recipe for epistemological collapse.
You've been waiting to use that phrase.
I have, and I'm not apologizing. But it's the right term here — when you can't know what's true, you stop trying, and that's when authoritarianism really wins.
There's a self-defeating loop possible here. You flood the zone with synthetic dissent, the regime cracks down harder because it can point to foreign manipulation, and the actual organic dissent gets delegitimized along with the fakes.
We've seen versions of this before. The Russian Internet Research Agency didn't just pump out pro-Trump content in 2016 — it also organized real protests and counter-protests, sometimes on the same day, on the same street, just to create chaos and make Americans distrust every political gathering. The goal wasn't to elect one candidate — it was to make the whole system feel rigged and illegitimate. If the Mossad's influence branch is operating on similar principles, even with different objectives, the collateral damage to the information environment is the same.
Which brings us to the verification problem. How do we even assess whether any of this is true? The Israel Hayom piece cites unnamed security sources. The case study is a photograph that may or may not have been released by Mossad — Ghasemi did resign, but we don't have independent confirmation of the causal chain.
We know Ghasemi resigned in November 2022, during the protests. We know there was an embarrassing photograph circulating. But the claim that the Mossad held it since 2011 and strategically released it — that's sourced entirely to Israeli intelligence officials speaking through a friendly newspaper. There's no second-source confirmation, no leaked document, no defector testimony. It could be true, it could be partially true, it could be a complete fabrication designed to make the Mossad look more capable than it is.
The photograph could have been released by Ghasemi's domestic political enemies. It could have been released by a genuine Iranian dissident. The Mossad might be taking credit for something they had nothing to do with.
Intelligence agencies do that all the time. Claiming credit for coincidences is practically a core competency. If an Iranian official resigns for internal reasons, and you put out a story six months later saying you orchestrated it, who's going to contradict you? The official can't say "actually I resigned because of a factional power struggle" without admitting the regime is fractious. The regime can't say "the Mossad is lying, we took him down ourselves" without looking unstable. You get the reputational boost either way.
The perfect crime — one where the victim can't report it without making things worse.
That's why I read the Israel Hayom piece as at least partially an influence operation in its own right. The article is the product. The revelation of the branch is designed to do psychological work on multiple audiences simultaneously. To the Iranian regime: we have penetrated your society more deeply than you knew, and we have patience you can't match. To the Iranian public: the regime's enforcers are hypocrites, and we're helping expose them. To Israel's allies: our capabilities are sophisticated and our resolve is long-term. To Israel's domestic audience: the government is prosecuting the Iran threat creatively, not just through costly military strikes.
Four audiences, one newspaper article. That's efficient.
It's what influence operations look like in practice. They're not usually one dramatic event — they're cumulative, multi-layered, designed to shift perceptions incrementally across different populations. The Ghasemi case is the headline-grabber, but the real work is the day-to-day mapping of Iranian society, the cultivation of media contacts, the slow building of networks that can be activated when needed.
Let me push on something. The Israel Hayom piece frames this in almost messianic terms — "we are not done with Iran, we are just getting started," the branch built to help bring down the regime. How much of that is genuine capability and how much is wishful thinking? We've been hearing "the Iranian regime is about to fall" for decades.
That's the essential question. The regime has survived massive protests in 2009, 2019, and 2022. It's survived economic collapse, international isolation, the assassination of top commanders and scientists, and sustained Israeli sabotage campaigns. It's brittle in some ways but remarkably resilient in others — particularly in its willingness to use violence against its own population. The idea that influence operations plus kinetic strikes equals regime collapse assumes a linear relationship between pressure and regime stability that history doesn't really support.
Authoritarian regimes don't usually collapse from external pressure plus internal dissent. They collapse when the security services decide they're on the losing side and switch allegiances. Everything else is just creating conditions.
That's actually the smartest part of this strategy, if the reporting is accurate. The influence operations aren't aimed at the general public primarily — they're aimed at elite cohesion. The Ghasemi takedown doesn't just embarrass one official, it sends a message to every other regime insider that the Mossad has files on them too, that their secrets aren't safe, that their position is precarious. If you can make regime elites paranoid enough about each other, paranoid enough about what might come out next, you might trigger the kind of internal fracture that actually does bring down a regime.
Paranoia as a service.
And it's not a new idea — the KGB was masterful at this during the Cold War, feeding misinformation designed to make Western intelligence agencies suspect each other, suspect their own officials. What's new is the technological toolkit — AI-generated personas, coordinated social-media amplification, the ability to micro-target different demographics with different messages simultaneously.
Let's talk about how this differs from what we normally associate with Mossad. The popular image is assassinations, sabotage, dramatic exfiltrations. The Fakhrizadeh hit, the Natanz explosion, the nuclear archive heist in 2018 where they stole literal tons of documents from a warehouse in Tehran.
Those kinetic operations are real and significant. But they're also costly, risky, and each one raises the stakes. Every assassination invites retaliation. Every sabotage operation risks escalation. Influence operations, in theory, operate below the threshold of war. You're not killing anyone, you're not destroying infrastructure — you're manipulating information. The Iranians can't exactly launch missiles at Tel Aviv because a photograph of a former minister got leaked.
No, but they can launch missiles at Tel Aviv because you've been assassinating their scientists for a decade and they need to respond to something. The influence operations don't replace the kinetic ones — they're a complementary track.
That's exactly how the Israel Hayom piece describes the doctrine — influence operations combined with kinetic strikes, not instead of them. The theory is that each amplifies the other. A sabotage operation at a nuclear facility is more demoralizing if there's a coordinated information campaign framing it as evidence the regime can't protect its most sensitive sites. An embarrassing leak about an official is more damaging if it lands during a protest wave that kinetic operations helped trigger by degrading the security apparatus.
The bombs and the narratives working the same target from different angles.
This is where the 2021 establishment date is interesting. David Barnea took over as Mossad director in June 2021. The influence branch was reportedly one of his first major institutional innovations. That suggests he looked at the Iran portfolio and concluded the existing toolkit — HUMINT, sabotage, targeted killings — wasn't sufficient for the objective, which appears to be nothing less than regime change.
Barnea's been fairly explicit about that objective, hasn't he? He's given speeches framing the Iranian regime as fundamentally irreformable, something that needs to be removed rather than negotiated with.
And that's a significant shift from the previous consensus, which was more about containment and delay — slow the nuclear program, degrade specific capabilities, buy time for diplomacy or for the regime to moderate on its own. Barnea's approach, if this reporting is accurate, is more ambitious and more controversial — it treats the regime itself as the target, not just its nuclear program or its proxy networks.
Which raises the question of whether Israeli intelligence has the capacity to actually deliver on that ambition. Mapping Iranian society is not a small project. Iran has eighty-five million people, multiple ethnic groups, complex internal politics, a sophisticated security apparatus of its own. The idea that a foreign intelligence service can understand it well enough to manipulate its fault lines — that's a big claim.
It's an enormous claim, and we should be skeptical of it. The CIA spent decades trying to understand and influence Soviet society with vastly more resources, and the consensus among historians is that their record was mixed at best. The KGB was much better at understanding the West than the CIA was at understanding the Soviet Union, partly because open societies are easier to study than closed ones. Iran is a closed society with a sophisticated counterintelligence apparatus. The Mossad undoubtedly has capabilities, but if the Israel Hayom piece makes it sound like they have a perfect real-time map of Iranian public opinion, that's almost certainly an exaggeration.
The IRGC isn't sitting still. They have their own cyber capabilities, their own influence operations, their own AI researchers. This isn't a one-sided game.
The Iranians have been running influence operations for decades — through cultural centers, through proxy media outlets, through social-media networks that push anti-Western and anti-Israeli narratives. They're not amateurs at this. The Mossad's new branch is entering a contested space against an adversary that understands it well.
We've got two intelligence services, both running influence operations, both using AI-generated content, both trying to manipulate the same populations. The Iranian public is caught in a crossfire of synthetic realities.
That's the tragedy at the heart of this. The actual Iranian dissidents — the real ones, the ones risking their lives for genuine political change — they get caught between two propaganda machines. The regime calls them Mossad agents. The Mossad's AI personas dilute their authentic voices. Everyone ends up less trusted, more isolated, more vulnerable.
There's a line in the Israel Hayom piece about the branch's doctrine — something about how influence operations aim to "weaken the Islamic Republic psychologically, politically, and socially." But psychological weakening isn't the same as liberation. You can break a society's faith in its institutions without building anything to replace them.
We've seen what that looks like in other contexts. Iraq after 2003 — the regime fell, but what followed wasn't democracy, it was chaos, sectarian violence, and eventually a new form of authoritarianism. Libya after 2011 — the regime fell, and the country became a failed state and a slave market. Regime collapse is not the same as regime replacement, and the influence-operations toolkit is very good at the first and useless at the second.
The Mossad can make the Islamic Republic look weak and hypocritical. It probably is weak and hypocritical. But what comes after is not something an intelligence agency can design.
That's the strategic risk. If the influence operations are successful — if they actually contribute to regime collapse — Israel inherits whatever comes next. A fragmented Iran with IRGC remnants controlling parts of the country, with ethnic separatist movements in the periphery, with no clear transition plan — that's not necessarily more stable or less threatening than the current regime. It might be worse.
Chaos has a way of producing things that make the previous regime look restrained.
I think about the Taliban in Afghanistan. The US spent two decades trying to build a democratic alternative, and the moment they left, the old regime's enemies swept back in. The Iranian opposition is real and brave and deserves support, but it's not clear it has the organizational capacity to govern an eighty-five-million-person country the day after a collapse.
Let me pull us back to the concrete. We've got a newspaper report in Israel Hayom, late May 2026, describing a secret Mossad influence-operations branch established in 2021. We've got one case study — the Ghasemi photograph, allegedly held since 2011 and released during the 2022 protests. We've got claims about AI-generated personas, coordinated social-media campaigns, and a doctrine of combining influence operations with kinetic strikes to bring down the Iranian regime. What should a smart listener take away from this?
I think there are three levels to read this on. Level one: the factual claims. Some of them are probably true — it would be surprising if the Mossad didn't have an influence-operations capability, and the Ghasemi case is plausible. Some are probably exaggerated — the omniscient mapping of Iranian society, the precision of the effects. And some might be entirely fabricated or misattributed.
Level two: the disclosure itself is an operation. The article is doing work regardless of whether every detail is true. It's signaling capability, sowing paranoia, shaping perceptions among multiple audiences simultaneously.
And level three: this is a window into how twenty-first-century intelligence agencies are evolving. The old model — recruit spies, steal secrets, occasionally blow things up — is being supplemented by something more ambient, more persistent, and in some ways more dangerous. Information warfare isn't a side project anymore. It's becoming the main effort.
The spies are still there, the bombs are still there, but the battlefield is increasingly what people believe.
That battlefield has no borders, no rules of engagement, and no clear way to measure victory. The Mossad can claim the Ghasemi operation as a success — he did resign. But can they prove the influence branch is actually weakening the regime in measurable ways? Can they prove the AI-generated personas are shifting Iranian public opinion rather than just adding noise to an already chaotic information environment? Influence operations are notoriously difficult to evaluate, and the people running them have every incentive to overstate their effectiveness.
Nobody ever got a budget increase by saying "our influence is ambiguous and our metrics are inconclusive.
And that's a problem across all intelligence agencies, not just Mossad. The CIA has the same issue, MI6 has the same issue. Influence operations are inherently hard to measure because you're trying to isolate the effect of your intervention from all the other factors shaping a society's political trajectory. Did the protests grow because of your social-media campaign, or because the economy got worse, or because a viral video of police brutality sparked outrage? You can't run a controlled experiment on a country.
You get credit for the rain if you're holding an umbrella.
That's the entire business model.
What's the most interesting unanswered question here, for you?
The AI personas. The Israel Hayom piece mentions them almost as an afterthought, but if the Mossad has operationalized language models for creating synthetic Persian-language influencers at scale, that's a capability that changes the nature of information warfare. And we have no independent reporting on it — just that one mention in a single article. I want to know how sophisticated these personas are, how they're deployed, whether they're detectable by current counter-disinformation tools. That's the story I'll be watching for.
For me, it's the archive. The Ghasemi photograph sat in a drawer for eleven years. What else is in that drawer? And when does it come out? If I'm an Iranian official with a complicated personal life, I'm not sleeping well after this article.
Which is, of course, the point.
The point is always the point.
And now: Hilbert's daily fun fact.
Hilbert: In the 1810s, explorers crossing the Gobi Desert discovered that when lightning strikes sand, it sometimes forms glass tubes called fulgurite — essentially fossilized lightning. What fascinated naturalists at the time was that these glass formations often contained traces of plant roots, suggesting the lightning had somehow preserved an underground relationship between silica and vegetation that would otherwise have vanished without a trace.
That's unsettling.
This has been My Weird Prompts. Our producer is Hilbert Flumingtop. If you want more episodes, you can find us at myweirdprompts.com. We'll be back next week.
