It starts like a modern joke, but unfortunately does not end with a punchline. Somewhere on Facebook, a video is playing, a seriously looking journalist announces the coup d’état in France with practiced drama. Tanks, takeover of power, a colonel against Emmanuel Macron, Paris in a state of emergency. The small problem: none of it happened. Nothing at all. No coup, no colonel, no upheaval. Just an idea, a prompt, a click – and millions of people watching.
One could laugh about it, if it weren’t so unpleasant. Because this “journalist” does not exist any more than the coup itself. She is a digital phantom, created by artificial intelligence, conceived by a man from Burkina Faso who sees himself as both a learner and a trainer in matters of AI. A man who later insists that he “did not mean any harm.” That is the sentence that now reliably comes up whenever the damage has already been done.
The story reads like an instruction manual for the post-factual age. The French president himself becomes aware of it because an African counterpart asks anxiously whether the republic in France has just collapsed. Facebook is asked to remove the video. Facebook reviews. Facebook decides: no violation of rules. Reality, it seems, has no fixed place in the terms of service.
The creator of the videos operates under the inconspicuous name “Islam,” has been showing several AI-generated clips since October, and is gradually testing the boundaries of reality. First men turning into snakes. Then rain made of whales. Then dragons above the clouds. One could have become suspicious at the latest with lion attacks in Paris, but apparently the rule is: the more absurd the world already seems, the less noticeable the nonsense is.
At some point, it becomes political. A real coup attempt in Benin provides the perfect opportunity; shortly thereafter, an AI-generated RFI journalist “reports” on takeovers, first in West Africa, then directly in France. Millions of clicks. Reach. Attention. And of course the subtle note on the side: Anyone who wants to create such viral videos themselves can feel free to get in touch via WhatsApp. Nowadays, they call this an educational campaign.
The truly fascinating aspect of this affair is not the technical achievement. That is banal. AI can now speak, look, and emphasize convincingly realistically. What is really remarkable is the moral emptiness accompanying this technology. A coup d’état is invented just like a headline used to be on the schoolyard. And when there is trouble, the retreat follows into the vocabulary of remorse. Fictional, poorly formulated, misunderstanding. Oops.
The man apologizes. Sincerely, as he emphasizes. He respects institutions, peoples, authorities. In the future, he wants to work responsibly, verify, improve. You almost believe him – just as you believe children who explain that they only wanted to quickly see what happens when you light a firecracker. The only problem is: On social networks, it’s not the firecracker that explodes, but trust.
Particularly striking is the detail that some of the videos are already self-reflective. An AI journalist explains there that she is not real at all, but artificially created – and in the same breath offers a course on how to produce exactly such deceptions. Transparency as a selling point. That is roughly like a pickpocket leaving a business card after the theft with a note about his next seminar.
Of course, all this could be dismissed as a bizarre footnote. A single hobbyist, a few videos, a misunderstanding. But the mechanism is familiar. Identities are forged, media brands imitated, political crises simulated. The pro-Russian disinformation campaigns have long perfected this. The only new thing is convenience. In the past, it took editorial offices, servers, structures. Today a laptop and a bit of boldness suffice.
Facebook, which did not want to remove these videos despite indications, plays the role of a bored doorman. As long as the rules are formally followed, nonsense is allowed in. Whether someone outside panics is only of minor concern. After all, the platform is not a Ministry of Truth. It is a marketplace – and panic sells well.
In the end, there are the “sincere apologies” and the deleted videos. The damage, invisible as it is, remains. Trust erodes not with a bang, but quietly, clip by clip. Eventually, you wonder with every message whether it’s real or just well animated. And that’s when everyone who still cares about reality has lost.
Not everything you can do should also be done. This sentence sounds old-fashioned, almost prudish. It comes from a time when technical possibilities were still accompanied by conscience. Today, the opposite often applies: what brings clicks is tried out. The apology comes later. Or never.
Perhaps the true irony of this story lies in the fact that the invented coup appeared more credible than the subsequent remorse. That is not a good sign. For anyone.
A commentary by C. Hatty