AI Disinformation: Viral Fake War Images

Massive Propagation of Disinformation by AI

In the first two weeks of the conflict between the United States, Israel, and Iran, the New York Times identified over 110 images and videos generated by artificial intelligence that reached millions of viewers. According to the analytics firm Cyabra, the majority of this content serves as pro-Iranian propaganda. These fake visuals, circulated on platforms like X, TikTok, and Facebook, include fictional explosions in Tel Aviv, supposedly attacked streets, and demonstrations of non-existent soldiers. Modern AI tools allow almost anyone to create convincing war simulations at little cost.

One particularly viral video showed missile strikes hitting the skyline of Tel Aviv. Another example involves the USS Abraham Lincoln: after the Iranian Revolutionary Guards claimed to have attacked the aircraft carrier, AI-generated images of a burning ship circulated on social media. The United States stated that the attack had failed and that the ship had not been damaged. President Trump accused Iran of using AI as a "weapon of disinformation" and accused Western media of "close coordination" with Tehran. The FCC Chairman, Brendan Carr, even threatened broadcasters with license revocation due to their coverage of the war.

Media researcher Marc Owen Jones from Northwestern University claims that the goal of this disinformation is to exaggerate Iran's military strength and make the war appear much more devastating for America's allies than it actually is. Real combat footage is typically filmed from a distance and at night, but the Times notes that the AI-generated fakes resemble Hollywood action movies, complete with mushroom clouds and bright hypersonic missiles.

Restrictions on Satellite Images

Satellite providers, such as Planet Labs, which operates the largest fleet of Earth observation satellites in the world, have extended the delay for releasing images from four days to two weeks, complicating independent verification. This decision covers all of Iran, allied military bases, and Gulf states. Vantor, formerly Maxar, has also restricted access to images of U.S. and allied bases. According to the Washington Post, these companies are not acting at the government's behest.

Critics argue that these restrictions could influence public narratives about the conflict and reduce transparency regarding attacks on U.S. bases. Gaps in real-time surveillance also facilitate the uncontrolled spread of disinformation and AI-generated content. For example, the Tehran Times published a satellite image claiming to show the destruction of a U.S. radar facility in Qatar, which turned out to be an AI-manipulated image from Google Earth. Satellite images are often used as evidence of successful strikes.

Impact on Traditional Media

German media outlets, such as Der Spiegel, Zeit, Suddeutsche Zeitung, WDR, Stern, and others, have been forced to withdraw AI-generated images from their publications after discovering they originated from Iranian networks. These images had been disseminated by the SalamPix agency and integrated into databases via the French agency Abaca Press. An Iranian photographer admitted to feeding images from a platform of the Iranian Revolutionary Guards into the supply chain without labeling them as such.

Consequences for Fact-Checking

OSINT analyst Tal Hagin summarized the dilemma with AFP: in the fog of war, it is difficult to assess the success of enemy attacks. OSINT, which relies on publicly available satellite images, had become a critical tool for circumventing censorship in countries like Iran. However, this trust is now being exploited by disinformation actors. Fake OSINT accounts are appearing on social media, disseminating AI-generated satellite images as genuine intelligence, undermining the work of legitimate investigators.