Trump Accuses Iran of AI Disinformation as Researchers Count 110 War Fakes

President Donald Trump on Sunday accused Iran of using artificial intelligence as a "disinformation weapon," claiming Tehran fabricated images of kamikaze boats, a successful strike on the USS Abraham Lincoln, and a 250,000-person rally that "never took place." The New York Times separately identified more than 110 unique AI-generated images and videos about the war in two weeks, a pace that Marc Owen Jones, an associate professor of media analytics at Northwestern University in Qatar, said far exceeds anything seen in prior conflicts. The accusations arrived the same day FCC Chairman Brendan Carr threatened to pull broadcaster licenses over war coverage Trump considered unfair.

Which reports, specifically? He never said. Reuters, for its part, reviewed footage shot at Basra's port. It showed explosive-laden Iranian boats ramming two fuel tankers, and at least one crew member was killed in the attack. Iranian state media did claim a hit on the Abraham Lincoln. Western outlets mostly didn't pick it up.

Both sides generate fakes

Iran's state media apparatus, emboldened by how far its content travels before anyone can fact-check it, has been pushing fabricated satellite imagery and inflated casualty figures since fighting started on February 28. It has also staged military victories that never happened. Tehran Times, a state-aligned English daily, posted a "before vs. after" satellite image on X claiming to show "completely destroyed" US radar equipment at a base in Qatar. Open-source intelligence researchers traced it to an AI-manipulated Google Earth image of a US base in Bahrain, not Qatar. Same parking lot, same cars. That's the tell. AFP detected a SynthID watermark, Google's invisible tag for AI-generated content. The manipulated photo still reached millions of views across multiple languages.

Iran's IRGC news agency Tasnim went further. On March 3, it published a number: 650 US military personnel killed in the war's first 48 hours. The Pentagon's count at that point stood at six. As of March 13, US Central Command's total stood at 13.

But fabrication runs in both directions. On March 4, the White House posted a video on its official X account merging real missile strike footage with clips from the Call of Duty video game. A choppy voiceover declared, "We're winning this fight." Five days into the war. The following day, another White House video celebrated "justice the American way" with scenes spliced from Braveheart, Breaking Bad, and Gladiator. The war has killed more than 1,300 people in Iran, according to officials there, and 13 US service members, according to Central Command.

"We have reached a level of realism in video, audio, and image deepfakes that for most people, it is not discernible from fact," Rumman Chowdhury, former head of ethics at X, told Rolling Stone.

When real photos become suspect

The volume of fakes has started corroding trust in authentic images, too, creating a fog where real and fabricated sit side by side in the same feed and look identical. Political scientist Steven Feldstein describes a growing category he calls "shallowfakes," content that manipulates subtly rather than fabricating entirely. A real photograph of an Iraqi airport showing smoke over a US military base was altered on March 1 with AI to replace the smoke with a giant fireball. The original was authentic. The alteration was minor. But the impression it left was not.

The New York Times ended up defending one of its own photographs this month. An organization called the Empirical Research and Forecasting Institute accused the paper of digitally manipulating a Tehran crowd image. "This is a genuine image, taken by a journalist in Iran on Monday, March 9, 2026," the Times wrote in a public statement. The accusation, it added, was "fundamentally flawed" and "dishonestly based on a re-posted version."

On March 3, Iranian state media said a strike on a school near a military base killed more than 160 children and staff. State outlets published an aerial photograph of a mass funeral alongside the report. Critics jumped on it, calling the funeral image AI-generated. BBC journalists geolocated the cemetery to a site 3.7 kilometers from the school, matching trees, road layout, and a nearby building against satellite imagery. Freshly dug graves appeared on satellite photos from the day after the funeral. The day before, bare ground.

Mahsa Alimardani at the human rights group Witness told the BBC she sees a split screen. Hold two truths at once, she urged. The Islamic Republic destroyed protest footage in 2022, hid prisoner deaths for years, scrubbed hospital records after crackdowns. Yet during this war, Tehran has been documenting civilian casualties with real resources. That documentation serves state propaganda. It is not automatically false.

That fog, where nothing looks trustworthy and everything looks the same, is exactly the environment social media platforms were supposed to manage.

X struggles to police its own platform

Elon Musk's X announced last week that creators posting AI-generated content about armed conflict without disclosure would lose revenue-sharing eligibility for 90 days. A defensive move for a platform broadly criticized as permissive toward disinformation since Musk's $44 billion acquisition in 2022.

Enforcement has been thin. "The feeds I monitor are still flooded with AI-generated content about the war," Joe Bodnar of the Institute for Strategic Dialogue told AFP. One premium account shared an AI clip depicting an Iranian nuclear strike on Israel. It gathered more views than the policy announcement itself. By a wide margin.

X's own chatbot compounded the problem. When disinformation analyst Tal Hagin asked Grok to verify a video of Iranian missiles supposedly striking Tel Aviv, originally posted by an Iranian state outlet, the chatbot repeatedly misidentified the location and date. A study by the Digital Democracy Institute of the Americas found that more than 90 percent of X's Community Notes never get published, leaving the platform's primary fact-checking mechanism largely inert.

Premium accounts with purchasable blue checkmarks have spread much of the AI content. One posted an AI video of Dubai's Burj Khalifa engulfed in flames, ignored a direct request from X's head of product to label it, and watched the post accumulate more than two million views.

"X's policy is a reasonable countermeasure," said Alexios Mantzarlis, director of the Security, Trust, and Safety Initiative at Cornell Tech. "It is unlikely that X will be able to guarantee both high precision and high recall for this policy."

A conflict saturated with fabrication

A Cyabra study of online activity found that the majority of AI-generated war videos push pro-Iranian narratives, often to falsely demonstrate military superiority. Nobody controls it anymore. Imposter accounts on Instagram, Threads, and X now pose as credible open-source intelligence analysts. They post fabricated satellite imagery with gibberish coordinates, poisoning the verification pipeline that journalists and researchers depend on.

Jones put it plainly. "Even compared to when the Ukraine war broke out, things now are very different," he said. "We're probably seeing far more AI-related content now than we ever have before."

For civilians in Iran trying to determine whether a strike report is genuine or whether an evacuation order is real, the fog carries weight beyond media criticism. The platforms carrying this content have not earned your trust. The governments producing it haven't either. New fakes land faster than anyone can flag the old ones. The verified photos sit three posts down, looking exactly the same.

Recommended

California Probes xAI Over Grok's 6,700 Sexualized Images Pe

Grok produced 6,700 sexualized images per hour while other deepfake sites averaged 79. California's AG opened an investigation the same day Musk claimed he saw "literally zero" illegal content. xAI's fix: charge $8 to keep using the feature.

Implicator.ai

Google Charges to Remove AI Watermarks It Calls Critical

Google's new image generator embeds "critical" authenticity watermarks in every AI creation. But pay $19.99/month and the visible mark disappears. The company monetizes the transparency tools it says protect public trust, while actual deepfake threats go unaddressed.

Implicator.ai

France Raids X Offices, Summons Musk as EU Opens Grok Probes

France raided X's Paris offices with Europol on Tuesday and summoned Musk for questioning. Hours later, the UK and EU opened separate probes into Grok deepfakes.

Implicator.ai