AI Propaganda and Polarization: More Reasons to Distrust the Media

Opinion From the Center

A recent slew of disinformation surrounding US-Israeli strikes against Iran has accumulated millions of views on X. AI generated videos of missile strikes in Tel-Aviv, children running for their lives, and “Chinese assisted Iranian” fighter jet attacks are the latest viral sensations.

The conflict in the Middle East has proliferated on social media, as X accounts spread fake headlines and false breaking news, while Iranian news outlets have shared false satellite imagery to exaggerate the severity of their retaliation.

However, this type of irresponsible reporting is not purely reserved for fringe X accounts or anti-Western regimes, as we have seen the same manipulation of photos and videos by our own influencers and media in the US, as well as by the Trump administration; often times with the intent of stoking division, confusion, and/or to build tribalistic political coalitions.

The Post-Truth Climax

The rise of artificial intelligence has sparked intensified discussions regarding its role in workplace efficiency and intellectual dishonesty, often sparking controversy amongst students, teachers, and professionals. But more recently, AI has become a common tool on social media, frequently being employed for lighthearted purposes, such as generating memes or humorous animal videos, but also being used for harmful ends, including the spread of internet disinformation.

The term “propaganda” dates back to Edward Bernays, the nephew of Sigmund Freud, who wrote the famous book, Propaganda. Bernays and other public opinion writers, such as Walter Lippmann, made the case that the public can be emotionally swayed by propaganda, and it is essential to “engineer consent” in order to align public opinion in the interest of business or government.

This discussion has shifted significantly from its 1920s roots, and now must account for social media, alternative media, and the ability for individuals to produce propaganda on their smart phone and with AI.

We have entered what many call the “Post-Truth” era – where truth was once objective, it's now become relativistic. A quote from Donald Trump’s character in the 2024 film, The Apprentice, says it best: 

“What is truth? What I say is truth, what you say is truth.”

Prior barriers to online disinformation, such as money, time, and legitimacy have all been eliminated, arguably by the normalization of AI content online.

The most severe affliction as a result of misinformation normalization is truth fatigue, where the truth is so scarce, so unclear, and often misunderstood, the public may lose interest entirely in finding the truth, and they become solely reliant on emotional impulses to guide their beliefs.

Division & Coalition Building

We all have likely seen AI in 2026 being used not just for writing college papers or editing photos, but for creating extremely divisive and offensive online content. This has been true of the internet for some time now. However, the normalization of AI content being distributed by political campaigns, corporations, or the White House is a new and unfamiliar animal.

The cost of using AI for divisive purposes was at one time a high cost – one that threatened legitimacy and credibility – especially for accredited news organizations or institutions.

But now, the mainstream use of AI to create false or offensive content has lowered expectations for professional and tenable reporting or posting. AI “meme” content is a new tactic being used by the Trump administration and other politicians to smear opponents and build coalitions.

One notable example was the recent post shared by Trump which depicted “former President Barack Obama and former first lady Michelle Obama's faces transposed on a monkey-like illustration,” reported USA Today (Lean Left bias), which Trump later deleted and condemned. The video portrayed many left-wing politicians and influencers on the bodies of other animals such as giraffes and donkeys, while Trump was the king of the jungle as his face was transposed onto a lion. 

The infamous “Trump Gaza” video, which Trump reposted on Truth Social, depicted a futuristic Gaza strip with golden statues of Trump, Elon Musk dancing surrounded by dollar bills, and Trump suntanning with Israeli Prime Minister, Benjamin Netanyahu – a bizarre post for any political professional.

The public association the administration has with divisive content like this has raised questions of Trump’s intentions. Does this type of content risk delegitimizing his administration? One answer is that this type of content spreads rapidly online because the emotional response of the viewer, either positive or negative, leads to an impulse to share the content.

Another answer may be that Trump is attempting to build coalitions by connecting groups that may disagree on certain issues but unite over others. Trump’s AI videos appeal to sects of the voters that have expressed a discontent with political correctness – a trend that many average voters can agree with, but is also popular amongst alt-right groups like Nick Fuentes’ Groyper movement or those who associate with white identitarianism. 

Furthermore, the futuristic utopia of the “Trump Gaza” video might build pro-Israel, anti-Hamas, and pro-tech sentiment amongst different groups, while the superimposed AI Obamas may be taken as a way to embolden and normalize racism while conflating it with disliking the Obamas.

Other politicians do this as well. A viral X post by California Gov. Gavin Newsom features Trump being WWE body slammed by Newsom, Barack Obama, Hakeem Jefferies, and others - purposed to generate anti-Trump agreement between liberals, leftists, and some on the Right who are discontented with Trump.

During the 2025 mayoral election race in New York City, former New York Gov. Andrew Cuomo posted an AI attack ad on Zohran Mamdani that depicted Mamdani “dressed as a socialist” for Halloween. These videos are not intended to be deep fakes, but rather low-cost smear campaigns.

The New York Times (Lean Left bias) published a report on the many times the Trump administration has utilized AI content to “attack enemies and rouse supporters.” Regardless of whether Trump, Newsom, Cuomo, or other politicians utilize AI content, the outcome is almost always the same.

AI content strays from traditional norms held by political figures, arguably delegitimizing parties and politicians. Certain content can reinforce racial or political stereotypes and crowd good faith disagreements.

Moreover, AI political meme culture can create exclusionary ideas about policy and social issues, painting one group as an enemy and others as allies by signaling in-group belonging.

These can create emotional coalitions driven by intuitive preferences that lack deep understanding of policy and ideology, taking advantage of the emotional impulses of the public by invoking feelings of utopia or outsider-fatigue.

Flat Out Lies

Manipulating the public’s emotions is one thing, but posting and reporting on fake videos and photos has been made even easier with AI.

In the wake of the death of Alex Pretti during ICE raids in Minnesota, a professional headshot of Pretti began circulating after MS NOW (Left bias) shared it on TV, their website, and on YouTube. The photo was an AI-enhanced version that depicted Pretti as “a few shades more tan than he actually was. It reshaped virtually every aspect of his face, including the angle of his jaw, the bend of his nose, and his hairline,” reported the New York Post (Lean Right bias). 

This dishonest reporting created what I call the Luigi Mangione Effect, which psychologists explain is a real phenomenon where attractive criminals manage to escape public blame.

A now-deleted post by the White House on X attempted similarly dishonest reporting. A photo of Nekima Levy Armstrong, a protester arrested during recent Minnesotan ICE protests labeled as “riots” by the administration, was posted by the White House that depicted her sobbing while being arrested. 

The photo was discovered to be AI-altered and the White House ultimately admitted to sharing the fake image. The AI photo was darkened and showed Armstrong sloppily crying with her face covered in tears. Opposite to MS NOW’s intentions, the administration attempted to disfigure the woman’s photo to encourage public blame of the suspect.

These are divisive, coalition building tactics just like the aforementioned AI meme content. The intention is to drive public opinion for or against a particular person or group. These deep fakes have not improved substantive conversation over whether Alex Pretti’s death was in vain, or if ICE raids/arrests are for just purposes, but rather have further polarized groups that have already been at odds. 

The success of certain media organizations or political groups relies on the deepening of polarization as their constituency feeds on it, primarily because polarization drives anger and resentment, and these emotions drive viewership and clicks.

The Future of Fake News

Social media has recently been bombarded with fake videos and images of the recent US-Israel-Iran conflict in the Middle East. While most legitimate news media has distinguished between the real and fake videos, non-corporate accounts with massive followings have been spreading disinformation. 

A popular Instagram account called @BestForces, known for its military content with nearly 700,000 followers, posted an AI generated video with the headline, “Brutal Footage Shows Iranian Ballistic Missiles Striking Tel Aviv.”

MS NOW, the White House, or other corporate/public institutions oftentimes must face public scrutiny when they post false content, however faceless instagram accounts do not play by the same rules. Influence on social media is not based on truth or legitimacy but on clicks & views, and the implications of alternative media misinformation are just as dire as mainstream misinformation.

Artificial intelligence has proven itself a tool for efficiency, writing, creating content, and so many more productive ways, however its more malicious uses have included emotional manipulation, deep fake proliferation, and widespread confusion and mistrust. Social media is a catalyst for this, but corporate media and political movements have proven themselves as equally untrustworthy.

The AI revolution does not show signs of slowing down, so we can expect this type of false information to stick around for a while. However, the change can be made in the way the public consumes their media. Our scrutiny of media must become more robust as the media’s ability to lie becomes more robust, and our trustworthiness of alternative news on TikTok, Instagram, and Facebook must be deeply reconsidered.

There is a lot of wisdom in the old adage: Believe half of what you see, and nothing that you hear.

Gabe James is a content intern for AllSides. He has a Center bias.

Reviewed by Editor-in-chief Henry A. Brechter (Center), News Editor Johnathon Held (Lean Right) and Research Assistant Emanuel Macuixtle (Left).