Since the birth of the internet, the world has inevitably changed, marking the transition from an analogue, locally rooted dimension to a global and constantly connected one. Its arrival has transformed the way we communicate and consume information, reshaped economic and working rhythms, and even altered the way we relate to one another. Over the years, however, it is not only society that has been reshaped. The web itself has changed so profoundly that its end can sometimes seem increasingly close.
Or at least that is what supporters of the Dead Internet Theory believe. The conspiracy theory emerged toward the end of the 2010s on American platforms such as 4Chan, before spreading to Reddit and YouTube. According to the theory, the web effectively “died” in 2016 because of the massive presence of bots across chats, forums and social media, designed to generate automated content and control the population. In 2021, the theory gained wider traction on the Macintosh Cafe on Agora Road forum, thanks to an anonymous user known as IlluminatiPirate, who argued in a lengthy post that algorithms had made digital spaces increasingly sterile. “It is important to make clear that the Dead Internet Theory is not an academic hypothesis,” explains Alessandro Tavecchio, a lecturer in Digital Journalism at SISSA in Trieste and an expert in science communication. “Rather, it is a colloquial term used to describe a system of gaslighting and propaganda on a global scale, based on the belief that Google, DARPA and similar organisations joined forces to fill the internet with bots and pollute online knowledge.”
It is no coincidence, however, that such a theory focused on the years 2016 and 2017. At the time, the web was going through a period of enormous expansion and social media was enjoying something of a golden age. But it was also the period when fake news, troll farms, which could simultaneously manage thousands of fake accounts and manipulate public opinion, and deepfake videos and images began to spread, exposing users to a much darker and less trustworthy side of the internet. Today, almost a decade later, the theory is circulating again, and this time it appears to have a more tangible basis, precisely because of the rise of chatbots and generative artificial intelligence. The problem is that the Dead Internet Theory, like any self-respecting conspiracy theory, originates from fears that are emotionally legitimate. The feeling that the web is becoming increasingly less human-centred, more alienating and stripped of meaningful parasocial relationships is real and understandable.
Added to this is the long phase of “enshittification”, the decline of platforms increasingly flooded with low-quality content that degrades digital spaces, bores users and relegates them to the role of passive spectators. TikTok’s feed offers a tangible example. Unlike Instagram or Facebook, the platform is not primarily based on interaction between people, but operates more like a large-scale, continuous entertainment engine. Besides pushing users to become detached consumers rather than prosumers, it is also home to a proliferation of accounts that create content exclusively with AI, reaching millions of views despite their nonsensical imagery, as in the well-known cases of Brainrot characters or AI-generated fruit. What fuels the conspiracy theory even further is that a large share of these interactions comes from automated accounts which, by generating traffic around a piece of content, eventually attract real users as well. On platforms such as X or LinkedIn, the process is even more visible. On the former Twitter, groups of bots comment on thousands of posts for the sole purpose of achieving virality and gaining access to monetisation features, while the world’s best-known professional networking platform is full of posts and job listings written in such an artificial and inauthentic tone that the absence of any human hand is immediately apparent. The impression, in short, is that only artificial voices are left talking online, inside an ecosystem where interaction between actual people is no longer essential.
A number of data points appear to reinforce this argument, including estimates suggesting that “a large share of web traffic and probably close to 50% of content written on social media is mediated in some way by AI or bots.” Even data recently released by Cloudflare points to a sharp increase in AI-agent activity, with bots accounting for around 57.5% of all requests to web pages, compared with 42.5% generated by humans. It is a historic crossover, arriving one year earlier than predicted by the American company’s own CEO, Matthew Prince, and marking another turning point for the future of the web. Then there is the now-familiar way social-media algorithms operate: people no longer explore freely and independently, because the content they see, the news they read and the influencers they discover are filtered through algorithmic systems calibrated around each individual user’s behaviour.
All these factors might therefore suggest that the Dead Internet Theory is actually correct, or almost a self-fulfilling prophecy. The reality, however, is that “it is closer to a ‘poetic’ diagnosis than to a genuine hypothesis, because it describes collective perceptions more than empirical or scientific evidence.” At present, artificial-intelligence systems still do not function without human input and, more generally, cannot independently conceive content or shape the online experience without the involvement of real people. Nevertheless, “we now have to accept that we are living in a post-AI era,” Tavecchio stresses. “It is an irreversible process, that much is true, and one from which the internet itself will emerge profoundly transformed. But it will not be a change that kills it, just as the rise of social media did not kill it in 2012.”
The professor points instead to a thoroughly human feeling: fear of the unknown, and of processes that people do not fully understand or control. After all, progress in this field has been both too extensive and too rapid to be easily absorbed. Until only a few years ago, bots could do little more than generate stock-market quotes or sports-news forecasts because those outputs were highly structured and predictable. Today, by contrast, extraordinarily powerful tools such as generative AI have reached such a vast audience that they are used on a daily basis.
“The only plausible prediction,” Tavecchio concludes, “is that people will increasingly favour virtual spaces resembling private conversation channels, such as Telegram and WhatsApp groups, Discord or Reddit servers, and even private Facebook groups, at the expense of more traditional platforms.”
The internet may therefore lose its nature as a giant, global and apparently transparent noticeboard, making way for a new era of smaller communities: more closed, harder to map, but experienced more concretely as social spaces made by humans.
Rachele Liuzzo