The AI mirage: How artificial intelligence fueled a war against Iran

An unverified AI-generated threat assessment triggered a preemptive US strike on Iran, exposing the dangers of outsourcing national security decisions to speculative algorithms and geopolitical manipulation, while revealing how intelligence distortion and technological overreach can escalate regional conflicts and undermine global stability. 

When US missiles slammed into Iranian nuclear facilities last weekend, American officials told the world it was a necessary strike to stop Tehran’s dash toward the bomb. But beneath the roar of the Tomahawks and the shockwaves of bunker-busting massive-ordnance penetrator lies a darker, more troubling story: one not of evidence, but of algorithmic speculation, of war justified not by hard intelligence but by the suppositions of artificial intelligence.

At the heart of this latest Middle East conflagration is Mosaic, an AI-driven program developed by Palantir Technologies and repurposed by the International Atomic Energy Agency (IAEA). Initially created for counterterrorism operations in Iraq and Afghanistan, Mosaic now scours over 400 million digital traces—tweets, movements, posts, logistics—to infer intent. It doesn’t prove what’s happening. It suggests what might happen. Much like Israel’s “Lavender” system, which flags Palestinians for drone strikes based on alleged associations with Hamas, Mosaic constructs a threat narrative out of statistical shadows.

Between June 6 and 12, Mosaic flagged an apparent surge of enriched uranium at Iranian facilities, including the Fordow Fuel Enrichment Plant (FFEP). The AI’s conclusion: Iran was weeks away from producing not just one, but potentially five nuclear bombs. This dramatic assertion made its way into IAEA reports—hailed by European allies and presented as a “dramatic document.” Yet it was based on inference, not verifiable facts.

Iranian and Russian officials dismissed the intelligence as a fabrication. Even Rafael Grossi, the IAEA’s Director General, admitted under pressure that there was no concrete evidence of a weapons program. The strike had already caused the damage, and its backers had successfully shaped the narrative. And that narrative, despite the intelligence community’s consensus to the contrary, became the justification for war.

Former US Congresswoman Tulsi Gabbard, now serving as Trump’s DNI, testified that there was no evidence Iran had pursued a nuclear weapon since Ayatollah Khamenei’s fatwa in 2002. Trump, however, rejected the assessment, relying instead on intelligence from outside sources, described as “managed, staged, and organised.” The fingerprints of Israeli influence, particularly through Mosaic-fed reports, were unmistakable.

OPINION: Pinprick sanctions and empty threats: Why Netanyahu isn’t losing sleep

The strikes themselves, carried out by B-2s and submarines using GBU-57 bunker-busters, caused limited physical damage. Iran had anticipated an attack and relocated much of its nuclear material. The symbolism, however, was unmistakable: a warning, a demonstration of American reach.

But the strategic calculus was flawed. Some US and Israeli strategists believed that strikes, coupled with drone infiltrations, would catalyse a colour revolution and the prospect of a colour revolution in Tehran, similar to early ambitions in Syria. Decapitate the military leadership, sow panic, and the regime collapses—that was the theory. Instead, it produced the opposite effect: a surge in national unity, public defiance, and regional coordination.

Iran’s foreign minister flew to Moscow the next day, a sign that Tehran, Moscow, and Beijing were recalibrating their axis. At the same time, Russia’s involvement in Ukraine limits its capacity for direct aid; China and even North Korea helped Iran build nuclear sites, says a defector, with its past assistance in centrifuge development, looms as a potential partner.

Meanwhile, Israel and the US are adjusting their strategy. Trucks disguised as civilian vehicles now deploy swarms of drones inside Iranian territory. Some are launched from Azerbaijan, while others are potentially launched by the MEK, which is believed to be coordinating with Israeli intelligence.

This hybrid war model—merging cyber warfare, AI-guided targeting, and covert drone operations—is exposing deep and dangerous flaws in long-held Western assumptions about regime change. The idea that limited strikes and technological superiority alone can spark political collapse is proving increasingly detached from reality. As members of the Five Eyes alliance scramble to reconfigure their national security frameworks to detect AI-generated misinformation, cyber-infiltration, and insurgent infiltration, it’s becoming clear that the age of traditional warfare is not just evolving—it’s rapidly becoming obsolete.

Tehran is reportedly studying Red Sea tactics to disrupt the Strait of Hormuz not through mining, but by slowing tanker traffic just enough to spike insurance premiums and roil markets.

This is a cautionary tale about overreliance on speculative intelligence. Grossi’s admission that there was “no concrete evidence” confirms critics’ fears: Mosaic didn’t detect a bomb. It projected a possibility, then sold it as inevitability.

This war, waged on the back of algorithms and speculative machine-driven logic, is far more than a reckless military gamble. It represents a profound shift in how wars are justified and fought, replacing verified facts with probabilistic assumptions and hard intelligence with digital guesswork. If the goal was deterrence, it has failed. If the hope was to spark a revolution, it has catastrophically backfired. And if this is the new face of warfare, then we are already losing the far more vital battle, for truth itself. 

READ: Israeli army purchases frozen as drones worth millions lost in Iran

The views expressed in this article belong to the author and do not necessarily reflect the editorial policy of Middle East Monitor.