49.6% of global internet traffic in 2023 was generated by bots. Only 50.4% by real humans. These are not made up numbers, but official data from theImperva Bad Bot Report 2024 by Thales. And now Google has decided to give a definitive push towards the dead internet with I see 3, its new AI video generator that produces content so convincing it looks like it was shot by professional directors. With synchronized audio, sound effects and dialogue that would make Hollywood envious.
The question arises spontaneously: are we witnessing the definitive death of the authentic internet? You know: it always takes me a while to answer.
Veo 3: When the Dead Internet Becomes Reality
Presented during the Google I/O 2025, I see 3 It's a quantum leap in AI video generation (and I can't wait to get my hands on it for hours at a time, but that's a personal thing). Unlike its predecessors that produced silent clips or clips with audio added later, this system from Google DeepMind generate video with native audio: synchronized dialogue, environmental effects, even conversations between characters that feel authentic.
The result is frightening in its perfection. Videos that show people talking, moving and interacting with such realism that it is almost impossible to distinguish between real and artificial. Available through the Google AI Ultra plan at 249,99 dollars a month (but not yet in Europe, at least not with “regular” channels), Veo 3 is no longer a laboratory experiment, but a commercial reality that is already flooding the web with synthetic content.
The theory of the dead internet, which until recently seemed like a paranoia of niche forums, is starting to look like a lucid diagnosis of our digital age. By the way: you know what I'm talking about, right?
Dead internet: from theory to documented reality
La dead internet theory It was born in online forums around 2021: argued that most web traffic was now dominated by bots and artificially generated content. Its supporters claimed that the “authentic” internet was dead between the 2016 and the 2017, replaced by an artificial ecosystem designed to manipulate public perception.
For years it was dismissed as a conspiracy theory. Then came the data (which does not confirm any conspiracy, but which somehow supports it).
Il Imperva 2025 Report revealed an even more disturbing truth: Automated traffic has surpassed human traffic for the first time in a decade, making up 51% of all web traffic in 2024. “Bad” bots alone account for 37% of total traffic, growing steadily for the sixth consecutive year.
Timothy Shoup of the Copenhagen Institute for Futures Studies predicted in 2022 that by 2025-2030, 99% of online content could be generated by AI. His predictions are becoming reality much faster than expected.
The artificial ecosystem that surrounds us
What makes Veo 3 particularly disturbing is not just its technical quality, but the way it fits into an already compromised ecosystem. As we have highlighted, the evolution of video AI has been dizzying: from grotesque experiments to content almost indistinguishable from reality in just two years. You know that video meme of Will Smith eating spaghetti? Here.
Meta has announced the introduction of autonomous AI accounts on Facebook and Instagram, profiles managed entirely by artificial intelligences that "will exist on our platforms, more or less like normal accounts", according to Connor Hayes, Meta's vice president of generative AI. They will have bios, profile photos, and will automatically generate content.
And the “Shrimp Jesus” phenomenon? bizarre AI images that get thousands of likes on Facebook? A tsunami. Indeed, only the tip of the iceberg of an ecosystem where the artificial generates engagement through the absurd.
When the authentic becomes the exception
Veo 3 accelerates this process exponentially. The videos generated by the system are not simple automatic clips: they are complete narratives with plot, characters and dialogue. The risk of disinformation is evident, but the problem goes beyond fake news.
We are witnessing the birth of an internet where authenticity becomes the exception, not the rule. Every video we watch could be artificial. Every online conversation could be with a bot. Every piece of content that moves us could be algorithmically designed to manipulate our reactions.
Studies show that humans can already distinguish AI content from human content solo 53% of the time: barely better than a coin flip. When even healthcare professionals correctly identify AI abstracts only 43% of the time, we understand that the problem goes beyond technical expertise.
The cruel irony of artificial engagement
Nanhi Singh, General Manager of Application Security at Imperva, said that “bots are one of the most pervasive and growing threats to any industry.” But the real paradox is that it is often the platforms themselves that indirectly encourage them.
The algorithm does not distinguish between authentic and artificial engagement: it counts clicks, likes, views. If a video generated by Veo 3 gets more interactions than one made by a human creator, the algorithm will promote it. The result? A vicious circle where the artificial thrives because it is more efficient at generating signals that the platforms interpret as “success.”
And soon, very soon the base of the pyramid will also be completely overwhelmed. of the creators.
Dead Internet and the End of Discovery
What makes the dead internet theory so disturbing is not just the amount of artificial content, but how it is changing the very nature of the online experience. The early internet was a place of unpredictable discovery: you could stumble upon a personal page, a niche blog, an amateur video that made you see the world from a new perspective.
Veo 3 and its ilk are replacing this unpredictability with a “streamlined version,” so to speak, of the human experience. Content designed to be more engaging, more viral, “more perfect” (pardon the term) than any human could ever create. But also infinitely emptier.
Human resistance
Is there hope? Perhaps. I say, indeed: certainly. Thomas Sommerer He called the shift to artificial content “an inevitable event,” but also noted that the dead internet theory reflects public perception rather than absolute scientific reality.
“Resistant” spaces are emerging: private newsletters (by the way, we are about to start our own, we already have 3000+ trusted subscribers, thanks), Discord closed, personal blogs (yes, the eternal return). Places where authenticity is more important than engagement, where human conversation is worth more than clicks. It's what some call "dark social": the exchange of content through WhatsApp, email, direct messages.
Even the platforms are reacting. Google has introduced SynthID, a watermarking system to identify AI content, although its effectiveness remains to be demonstrated on a large scale.
The future we have chosen
Veo 3 is not just a technological product: it is a symbol. It represents an internet where algorithmic efficiency has replaced human authenticity, where artificial perfection is preferred to real imperfection.
It’s not necessarily an apocalyptic future. But it’s certainly a different future than the one we imagined when Tim Berners-Lee designed the Web as an “open and accessible” space to connect humanity.
Today, as Veo 3 generates video indistinguishable from reality and bots dominate web traffic, we must ask: What happens when we can no longer distinguish the real from the artificial? When the internet becomes a stage where the main actors are no longer human?
The dead internet theory was wrong about one thing: the internet is not dead. It has transformed into something entirely different. And perhaps, for the first time, we should admit that we no longer know exactly what it has become.
