Pro-Iran networks have utilized culturally savvy memes to poke fun at President Trump and portray anti-American sentiment as heroic during the opening week of the Gulf War.
These were more successful than U.S. media outlets, which had traditionally attempted to convey messages using other forms of communication. A video depicting Donald Trump transforming into a crying infant while Iranian missiles fall from the sky; another video displaying an animated style of resistance fighter (similar to an Anime character), who is battling an American mech. These are no low-budget, poorly produced video clips created in some guy's basement. Rather, they represent part of the coordinated AI-propaganda campaign by the Institute for Strategic Dialogue (ISD). The videos received over 1 billion views on X within the first month of the conflict.
The sheer number of people viewing this propaganda represents a major turning point in cyber warfare. Traditionally, state-run propaganda campaigns involved bots posting tweets containing simple images or basic photo editing. In contrast, the most recent pro-Iran campaign involves using AI generated video-clip propaganda, which utilizes current Internet trends and subcultures. According to a study completed by The Economist, dated April 18th, the pro-Iran propaganda network is succeeding because their content is both humorous and culturally relevant to the intended audience. Additionally, the content is also technologically superior to anything previously seen in propaganda campaigns, thus appearing as if it was produced organically and therefore increased chances of being viewed and shared virally.
It can be seen that each video was researched carefully. For example, one series utilized the visual style of "sigma male edits," a type of image/video that became extremely popular among young males on TikTok and Instagram. However, instead of showing typical businessmen and professional athletes, this series depicted Hezbollah fighters. The second used Vaporwave aesthetic styles, such as bright colors and retro graphics, to depict American military equipment as old-fashioned relics. This particular clip garnered approximately 47 million views prior to it being removed by X; however, copies are still spreading rapidly throughout social media via Telegram.
It should be emphasized that what makes this campaign so effective is not solely dependent upon advanced technology. The creators of the content behind this campaign are familiar with the types of memes currently trending; they know how to create them timely, utilizing the correct memes to encourage sharing on Western-based social media platforms. Essentially, rather than translate propaganda into the vernacular of their target audience; they are generating it natively in the visual vernacular of that same audience.
The Institute for Strategic Dialogue tracked the way in which these videos spread through what they refer to as 'laundering networks'; accounts that seem legitimate and post content regarding sports, entertainment etc., followed immediately by posting large quantities of politically charged content at the time of great importance. This suggests that there was months worth of planning prior to activation of inactive accounts at the moment of need.
Typical counter-message efforts have proven unsuccessful in countering this strategy. The State Department's attempt to produce videos outlining U.S. policy positions or attempting to debunk false claims of anti-U.S. rhetoric is ultimately unable to compete with content designed to elicit an emotional reaction and subsequently generate additional shares. As noted by a researcher at the Institute for Strategic Dialogue, "they're not trying to sell you on facts... they're trying to make resistance cool and America weak."
Get the latest model rankings, product launches, and evaluation insights delivered to your inbox.
Technological capabilities vary greatly depending on the individual videos contained within this campaign. Some clearly display evidence of AI artifact; flicker or distortions temporally; inconsistency in lighting between frames; etc. Other videos contain blended AI-generated elements and authentic footage, where detection would require frame by frame analysis. According to reporting by The Economist at least twelve different AI video model types appear to be in use although exact platforms remain unknown.
Responses from X have varied. While high profile videos have been removed after gaining widespread popularity, numerous iterations (smaller versions) continue to circulate on social media. When asked for further information regarding detection methods or if X will develop new methods for detecting AI-generated propaganda content; X did not respond.
Iran has certainly tried digital influence operations before; however, the current operation is an order of magnitude greater. Iran used to run fake news websites and Twitter bots (from 2019-2023) pushing articles through their networks; even though these were consistent, few times did they achieve viral status outside of Persian language social media circles. The major difference was the use of artificial intelligence generated video content fluent in internet memes/culture, which changed everything.
Timing is important. As consumer grade AI video content options became accessible at a quality level equal to what took weeks of work and expensive VFX teams just two years ago, AI generated video content flooded the web. The democratization of creating video with consumer-grade AI tools gave both independent creators and government-run propaganda organizations the means to create sophisticated propaganda for the cost of $1,000s instead of $millions.
Cultural relevance is far more valuable than high-quality production values when it comes to getting viral reach. Most western platform detection systems are still based around detecting static images and text-based input - not video. Old ways of countering propaganda - traditional fact checking and counter messaging - simply don't work well against propaganda presented in an entertaining format.
As audiences become increasingly unable to tell if they are viewing actual footage or synthesized AI-generated footage, the "liar's dividend" will continue to grow.
At least at present, there are no public Western government responses to the AI Video Campaign. This is less a matter of technical capability than a conceptual one. How does a government respond to propaganda that looks nothing like propaganda? Propaganda that people share because they found it funny or visually appealing versus them being persuaded by its message?
We believe the next iteration is already underway. ISD researchers identify a group of newly created accounts posting AI-generates videos about Taiwan. While stylistic approaches appear similar, cultural references seem to differ. We cannot yet say whether these represent further expansion of Iranian interests or another actor adopting the Iranian strategy.