AI and Information Warfare: When War Arrives Dressed as Entertainment

This talk was prepared at the invitation of the EDI Global Forum at Fondazione Morra Greco, for a conversation with cultural institutions about artificial intelligence, information warfare, and the changing conditions under which publics interpret reality.

Building on my article « Quand la guerre informationnelle passe par le Lego, l’anime, le jeu, la satire et l’IA »I began from a strange and troubling object: AI-generated LEGO-style videos depicting real geopolitical conflict.

A children’s toy. A war. Millions of views.

At first glance, the format may seem almost absurd. But that absurdity is precisely where I wanted the analysis to begin. These videos are not simply “generated content.” They are not only memes. They are not classical propaganda posters translated into digital form. They are platform-native synthetic formats that frame conflict through narrative and affect: short, emotionally coded, culturally familiar, and built to make events instantly legible, memorable, and shareable.

My argument was simple: today’s information warfare does not move only through spectacular fakes. It also moves through everyday cultural objects, aesthetic forms, and familiar formats that quietly shape narratives, normalize positions, and re-engineer public perception.

Why I chose to take the LEGO videos seriously

I did not choose these videos because I wanted to condemn the format. Quite the opposite. I think the LEGO material needs to be read carefully, not dismissed too quickly.

Before becoming a professor in educational technology, I taught visual arts for fourteen years. So when I look at these videos, I do not only see “content.” I see composition, rhythm, contrast, emotional pacing, and visual strategy. And I have to say it plainly: the craft is impressive.

LEGO creates a powerful semiotic collision. On one side, we have childhood, play, modularity, innocence, familiarity. On the other, we have war, military force, geopolitical violence, destruction, and death. The collision between those two worlds is not accidental. It is doing political work.

The aesthetic does not simply soften violence. It re-embeds violence inside a familiar cultural form. It makes the viewer pause, laugh, recoil, admire, question, or forward, sometimes before deciding what they actually think.

That is why I argued that the unit of analysis is not only the post. It is the format.

A post can disappear. A format travels.

From facts to frames

One of the central ideas I explored in the talk was that the contemporary struggle is no longer only about establishing facts.

Facts still matter. Verification still matters. Journalism, libraries, archives, courts, peer review, and public institutions still matter.

But facts do not arrive alone. They arrive framed.

A fact can arrive as tragedy, humiliation, joke, victory, betrayal, or injustice. The frame often arrives before the audience has time to analyze the claim. In platform environments, the first battle is often not over whether something is true or false. It is over how the object will be received, felt, remembered, and shared.

This is where aesthetic forms become politically powerful.

The LEGO videos simplify without necessarily neutralizing. They make complex scenes immediately legible. They turn geopolitical actors into manipulable figurines. They make the powerful look playable. They increase memorability through contrast.

In other words, the format does argumentative work before the caption arrives.

Not all synthetic media are morally equivalent

A crucial distinction in the talk was that I am not putting LEGO satire, counter-narrative, harmful deepfakes, fabricated evidence, and synthetic impersonation in the same moral category.

They are not the same thing.

The LEGO videos I discussed are not deepfakes in the strict sense. They do not impersonate a real person saying or doing something they never said or did. But they are synthetic media: AI-generated cultural objects that stage political reality.

That distinction matters.

The danger begins when synthetic media impersonates real people, fabricates evidence, erases accountability, denies real suffering, or makes authentic harm easier to dismiss.

So the question cannot only be: is this fake?

The stronger question is: what is this synthetic object doing politically?

From deepfakes to synthetic warfare

Deepfakes are already part of contemporary conflict.

The 2022 deepfake of Volodymyr Zelensky instructing Ukrainian troops to lay down their arms was crude and quickly debunked. But the artifact failed while the precedent remained.

Since then, synthetic media has become faster, cheaper, more culturally coded, and harder to detect. We have seen fake explosions, voice cloning, synthetic campaign materials, AI-generated political imagery, fabricated atrocity claims, atrocity denial, battlefield deception, and wartime generative satire.

This is not a future problem. It is already an active layer of political communication.

One of the most dangerous effects is what Chesney and Citron call the liar’s dividend: once deepfakes are known to exist, powerful actors can dismiss authentic evidence as fabricated. The harm is no longer only the false video or the false audio. The harm is the doubt cast on everything else.

Every credible deepfake makes every inconvenient truth easier to deny.

Slopaganda and the system behind the visible content

In the talk, I also discussed slopaganda: generative propaganda produced quickly, cheaply, and at scale, optimized not necessarily to persuade through depth, but to saturate or orient the information environment.

The goal is not always agreement.

The goal is often attention’s terrain.

To crowd the space. To make truth fight for oxygen. To keep people reacting, reposting, doubting, laughing, arguing, or chasing fragments.

This is where computational propaganda remains useful, but also needs to be stretched. What I am seeing is not always top-down coordination from a single command center. It is often more diffuse: militant producers, para-institutional actors, platform incentives, relay accounts, and ordinary users who translate, remix, and recirculate content because they find it funny, moving, clever, or true.

No single author.

No single command.

A coordinated style.

That is why it is so difficult to govern.

Counter-visibility

I also wanted to complicate the discussion.

Not all of these formats are simply deception.

Some of them create counter-visibility.

By counter-visibility, I mean the act of forcing something into public perception when it would otherwise remain unseen, unfelt, or morally illegible. Some synthetic or memetic formats translate realities of the ground into popular language. They can make ignored violence visible. They can cross the threshold of indifference.

But counter-visibility is not automatically truth. It has frames. It has silences. It has politics.

So the distinction matters:

Counter-visibility is not truth.

But it is also not nothing.

If we collapse that distinction, we misunderstand the terrain.

Why I addressed cultural institutions

I ended the talk by turning directly to cultural institutions because I believe they are not peripheral to this issue.

• Museums teach the reading of objects.
• Libraries teach the reading of sources.
• Archives teach the reading of memory.
• Schools teach all three at once.

These are exactly the competencies needed now.

The public does not encounter synthetic media in a neutral state. People arrive already shaped by feeds saturated with AI-generated, semi-synthetic, satirical, affective, ambiguous, and platform-optimized content.

Detection matters. Provenance standards matter. Watermarking matters. Platform moderation matters.

But none of these replaces public interpretation.

The defence is technical and cultural. Mostly cultural.

Cultural institutions are among the few public spaces where people can still slow down, look carefully, compare, contextualize, ask better questions, and interpret images, narratives, memory, and power with others nearby.

That is not nostalgic.

It is strategic.

From “spot the fake” to “read the frame”

In the talk, I proposed five interpretive competencies that I believe should guide the next generation of media literacy work.

• First, read the form, not only the fact. What is the format doing before the caption arrives?
• Second, trace the chain of circulation. Who made it, who relayed it, who translated it, who benefits?
• Third, distinguish satire, propaganda, and counter-visibility. The same generative tool can serve different political functions.
• Fourth, name the affective register. Is the object mobilizing indignation, irony, grief, belonging?
• Fifth, ask who becomes visible, and who does not.

The model generates.

The ecosystem weaponizes.

Literacy lives between those two verbs.

The task ahead is not simply to detect synthetic media. It is to understand how synthetic objects move through platforms, how they attach to affect, how they borrow cultural forms, how they shape public memory, and how they reorganize the conditions under which reality becomes believable.

That is the challenge cultural institutions now face.

Not because they created the problem.

But because they remain essential to the defence of interpretation.