How AI Is Skewing Perception of the Iran Conflict

As the war with Iran intensifies, content made using artificial intelligence (AI) is emerging online, fueling narratives that are either misleading or outright fabrications.

Accounts tied to Iran have falsely claimed the Islamic regime has made successful strikes on U.S. naval ships, bases, and troops, using AI-generated images and video to convince the public that the claims are true. It is the latest example in an increasing trend of AI being used during wartime to spread false information to sway public opinion.

Experts say this AI-generated propaganda, which has been used to great effect in other recent conflicts and wars around the world, is an easy way for governments to try to shore up public confidence and power while demonizing the opposition.

Below are some examples of this type of content, as well as the reasons why they’re considered fabrications.

Fake Fallout From Iranian Strikes

In one such instance of AI-generated content, the Iranian Tehran Times alleged that Iran destroyed American radar equipment at a U.S. naval base in Qatar.

The outlet posted side-by-side satellite imagery in a Feb. 28 post on X with a “BREAKING” and “Before vs. after” caption.

The after picture showed damage to several buildings and appeared to be based on real satellite imagery of the Naval Support Activity base in Manama, Bahrain, from February 2025.

The Epoch Times confirmed the before image, a satellite photo, through Google Earth and observed several illogical discrepancies between the before and after photos: Buildings at the bottom of the images didn’t match, and three vehicles were parked in the same exact locations in both photos, despite the before and after shots allegedly being taken a year apart.

A program called SynthID, made by Google, can detect whether the company’s AI tools created or altered a video, photo, or audio file.

In the case of Iran claiming a successful strike on the U.S. naval base, SynthID found that the before and after images were generated or edited with Google AI tools.

The Tehran Times shared additional photos on March 6, claiming “destruction and fires” at a U.S. base in the United Arab Emirates (UAE).

While SynthID did not identify the photos as created or edited by Google AI tools, X’s built-in AI system, Grok, said the images showed “strong” signs of being AI-created or at least heavily edited. Grok could not find any matching posts or articles tied to real events, which was unusual for dramatic photos showing damage to an American base.

X’s AI system highlighted several reasons to believe the photos were AI-created or edited. Heavy rain appeared to be falling in one photo, which didn’t align with how rain, smoke, and fire would have interacted in an authentic event, Grok said.

“Overly cinematic composition, unnaturally uniform flame patterns and smoke density without real physics inconsistencies” were also noted as well as an absence of camera artifacts like grain, flare, or depth-of-field variance typical in authentic photos.

Deceit in Wartime

It’s not always clear where the false claims originate. They could come from groups tied to the government or seemingly random users who nonetheless fuel false narratives.

Another AI-generated post, from an account apparently not tied to the regime, depicted a massive explosion at an international airport in Irbil, Iraq, which was near a U.S. military base that Iran had been targeting.

The image, depicting an enormous fire and a volcano-like column of black smoke, appeared to be inspired by a real photo showing a much more modest plume of smoke over the airport. Visual inconsistencies in the image were found by SynthID, which said Google AI tools were used to generate or edit the photo.

The Russia–Ukraine war and Israel’s 12-day war with Iran last summer also included AI-generated fabrications, as noted by think tanks.

“The volume of AI content is starting to just pollute the information environment in these kinds of crisis settings to a really terrifying degree,” Melanie Smith, senior director of policy and research on information operations at the Institute for Strategic Dialogue, told The Associated Press. “The inability to get access to verified and credible information in times like this—it’s getting harder and harder to do that.”

In a 2023 report, the Center for Strategic & International Studies highlighted AI’s influence on the Israel–Hamas war, stating bot accounts used by Hamas-aligned actors were used to amplify the terrorist group’s narratives.

“The backing of Iran … and the reported involvement of China and Russia in amplifying these narratives indicate a broader trend of social media as a tool in proxy warfare and the shift in military strategy to incorporate communication campaigns,” the report said.

It’s unclear how much of this AI-generated misinformation is reaching the Iranian people, as the government has imposed an internet blackout. WITNESS, a group that uses video to document human rights abuses, indicated last year that AI-generated content, which it said increased during the 12-day war, created an acute problem for Iranians as they faced “information pollution.”

“Citizens are often unable to verify or cross-check information fed to them by state or conflict actors,” WITNESS authors Mahsa Alimardani and Sam Gregory wrote in an article for the Carnegie Endowment for International Peace.

That problem grows when Iranian state media fuels false narratives, including from the government, as it has in the current conflict. Online, sources may build on such government propaganda as well.

Iran’s Islamic Revolutionary Guard Corps (IRGC) previously said it struck an American aircraft carrier, the USS Abraham Lincoln, with ballistic missiles. Online accounts seemingly not tied to official Iranian sources then backed up the IRGC’s claims with fake footage.

“Iranian missiles have sunk USS Abraham Lincoln aircraft carrier in the Persian Gulf,” one such account said in a post on X, which has garnered more than 8 million views. A community note on the post clarified that the footage was actually from a military-themed video game.

U.S. Central Command responded to the claim on X.

“Iran’s IRGC claims to have struck USS Abraham Lincoln with ballistic missiles,” it wrote. “LIE.”

Combating Propaganda

While there have been rapid developments in AI in recent years, along with widespread access to AI generation, with social media allowing it to proliferate at unprecedented speed, technologies for mitigating false narratives are also rapidly developing.

X has Community Notes, where verified users can directly address posts’ claims or offer critical context or missing information. Any account on X can rate the community note as helpful or not, and why. These notes have frequently appeared on posts about the latest conflict in Iran.

Nikita Bier, head of product at X, said on March 3 that the platform was revising policies for its program that allows users to earn revenue.

Now, those who post AI videos of an armed conflict without a disclosure that it was made with AI will be suspended for 90 days from the platform’s Creator Revenue Sharing program. Bier wrote on X that any subsequent violation will result in permanent suspension.

Sarah Rogers, the State Department’s under secretary for public diplomacy, praised the move. “You don’t need a Ministry of Truth to incentivize truth online,” she wrote on X.

Google and Meta did not immediately respond to requests for comment on any further initiatives they’re undertaking to counter AI misinformation from Iran.

Experts say that people should be diligent and critical of the information they consume, especially as AI progresses to a point where it’s nearly impossible to tell if the photos and videos they create are real or fake.

They suggest finding corroborating sources, like a media outlet, independent journalists, government sources, academics, researchers, or even nonprofit organizations.

A watermark is a telltale sign of manipulation on a video or photo. Sometimes, these can be burned on by the AI generator that made them. However, these watermarks can be blurred or hidden to prevent a viewer from knowing the imagery was AI-generated, which is where a program like SynthID can be helpful.

Those technologies are limited, however, as Alimardani and Gregory said last year, they sometimes produced conflicting results.

“Sophisticated detection methods can yield conflicting findings, leaving it up to journalists, human rights defenders, and civilians in conflict zones to navigate the fog of war without a clear signal of what is real,” they wrote.

The Associated Press contributed to this report.