Thursday, July 10, 2025
[Web Culture] Is the Internet Overrun by Bots and AI?
A growing share of online content is generated by algorithms, blurring the line between human and artificial interactions.
▸ Origin and premise of the theory
The Dead Internet Theory was born on online forums around 2021 and popularized by an article in The Atlantic the same year. At the heart of this conspiracy theory is the idea that the web “died” around 2016-2017, meaning that the vast majority of online content is now produced by bots (automated programs) or AI, pushing humans into the background. According to its proponents, there is a coordinated effort (by governments or large corporations) to flood the Internet with artificial content in order to manipulate opinion and control behavior. In other words, we would increasingly be browsing an Internet where genuinely human interactions are a minority, drowned in a mass of fake engagement and algorithmically generated posts.
This theory may seem extreme (and it is, in its literal reading), but it rests on very real trends. For example, automated web traffic has exploded in recent years. A report by the cybersecurity firm Imperva revealed that 40.8% of Internet traffic in 2020 came from bots, including 25.6% from malicious bad bots (spammers, scrapers, hackers, and so on). That is a quantifiable fact feeding the idea of an Internet that is less and less “human.” In parallel, advances in artificial intelligence (chatbots and deepfakes in particular) make it possible to mass-produce text, images, or videos indistinguishable from the real thing. The Dead Internet Theory extrapolates these trends into a near future where the Internet would be entirely flooded with artificial content: a dystopian vision that, while unproven, pushes us to think about the signals already present today.
▸ An increasingly artificial web: recent examples
Several recent phenomena lend credibility (or at least context) to this theory by illustrating the rise of bots, sponsored content, and fakes of all kinds on our digital platforms.
Bots and fake accounts on social media
The most emblematic case is undoubtedly Twitter. For years, Twitter has faced the proliferation of bots and automated accounts that interact like fake users. In 2022, during Elon Musk's acquisition of Twitter, the question of how many accounts were fake became crucial: Musk publicly challenged Twitter's official estimate (fewer than 5% bots among monetizable active users) and cited third-party analyses suggesting that 11 to 14% of accounts might actually be bots. That episode highlighted the scale of the problem: if even a major social network struggles to count its fake users precisely, it means bots have taken a significant place there.
More recently, after Twitter rolled out a monetization model encouraging verified accounts to post viral content, an army of “smart” bots emerged. Bad actors connected language models (ChatGPT-style AI) to automated accounts to flood the reply threads of popular tweets, hoping to capture a share of the ad revenue tied to those posts. The result? Conversations where supposed users with blue badges are in fact just chatbots talking to each other: a genuine “battle royale” of bots, to borrow one tech journalist's phrase. We reach a point where it becomes fair to ask, while scrolling your feed: how many of these interactions come from real people, and how many are the work of machines?
Disguised ads and omnipresent sponsored content
Another face of this “fake” Internet shows up in the proliferation of sponsored content, native ads, and other pretenses within seemingly editorial content itself. Most large platforms and online media have integrated ads that mimic the form of an ordinary post or a classic news article, blurring the line between information and commercial promotion. According to a study from Boston University, fewer than 1 person in 10 can clearly tell a piece of journalism from sponsored content promoting a product or brand. These “native ads” are designed to blend into the news feed, with the same style and tone as real articles. An estimate cited by Forbes even predicted that native advertising would account for 75% of ad revenue in 2021: a sign that this format has taken over the media space.
Concretely, this means that in our Facebook, Instagram, and TikTok feeds, between two posts from friends or creators we follow, slips a multitude of content paid for by advertisers, sometimes presented by influencers or generated automatically. On Instagram, for example, thousands of viral recipe, tutorial, or challenge videos have flourished that turn out to be brand-sponsored, or made by virtual influencers. The consequence is an erosion of trust: if anything can be an ad that does not say its name, the reader/user develops a generalized distrust of what they see online. Raw, authentic information has a harder and harder time cutting through an ocean of messages driven by marketing interests.
Telling the authentic from the fabricated: a growing challenge
Finally, the rise of fake news and AI-doctored content makes telling true from false ever harder. A striking case occurred in May 2023: a doctored image of an explosion near the Pentagon, generated by artificial intelligence, circulated massively on social media. Shared even by verified accounts, this fake news briefly sowed panic, to the point of causing a slight dip in the stock market before being debunked. Disinformation experts confirmed that the image bore the “signature” of AI generation (inconsistent pixel-level details), but its lightning spread shows how much a convincing visual is enough to fool thousands of people, including official media players.
That is only one example among many: one could also mention the photo of the Pope in a luxury puffer jacket that fooled internet users (also AI-created), or those deepfake videos that put any words in the mouths of public figures. Even on Wikipedia, known for its community's vigilance, bots take part in the writing: the magazine Epsilon recently reported that 5% of Wikipedia's English-language articles are now written by bots. Moreover, a Europol report anticipates that by 2026, 90% of online content could be generated by AI: a dizzying figure which, if it holds true, would partly realize the fear of an Internet “full of fakes.” Without going that far, we already see the production of artificial text, images, and video industrializing: AI-written blog posts to feed SEO, fake profiles generating online reviews, automated comments under posts… Everything conspires to muddy the waters. Digital “authenticity” is in crisis, because we no longer have any certainty that what we read or see comes from a sincere human or a well-programmed robot.
▸ What impact for digital professionals?
Why should digital professionals (marketers, communicators, data analysts, AI experts, cybersecurity specialists…) care about these developments? Precisely because the quality of online information and user trust are at the heart of many digital jobs.
Marketing & Communication: if consumers are saturated with artificial or misleading content, it becomes harder for brands to build trust and authentic engagement. Digital marketing campaigns risk drowning in a flood of bot-generated noise, or being perceived as inauthentic themselves. Communicators must double down on transparency (clearly flagging sponsored content, for instance) and creativity to stand out from a digital hubbub full of fake accounts and AI. Paradoxically, the quest for authenticity could become a major asset again: users value content embodied by real people, with an identifiable voice, as opposed to formatted, automated communication.
Data & analytics: for data analysts, the pervasiveness of non-human traffic complicates the interpretation of metrics. How can you trust web or social analytics if a significant share of clicks, views, or comments comes from bots? You need to develop tools for detecting inauthentic activity (anomaly detection, anti-bot filters) to clean the datasets. Otherwise, business decisions could be biased by artificial signals (for example, a company might wrongly believe in the popularity of a product based on interactions inflated by bots). In the field of monitoring and intelligence, telling true from false becomes a major challenge in order not to be manipulated by automated influence campaigns.
Artificial intelligence: for AI and development professionals, the proliferation of generated content poses a twofold problem. On the one hand, ethically, we must establish guardrails (AI text/image detectors, digital watermarks, etc.) to avoid abuse and help the public identify what is artificial. On the other hand, there is the risk of the self-reinforcing loop: AIs are trained on web data, and if that data is itself largely generated by AI, we head toward qualitative impoverishment (a bit like a photocopy of a photocopy losing sharpness). Researchers at Oxford and Cambridge have warned that conversational models feeding on already-synthetic data could, over time, produce absurd or erroneous content that no longer “means anything.” AI engineers must therefore be extra vigilant about the quality of training datasets, and consider solutions (for example, including more verified human data, or developing AIs able to recognize generated content).
Cybersecurity: finally, for information security experts, the rise of bots and automated content is an expanded threat surface. Bad bots can run large-scale phishing attacks by personalizing their messages with AI, generate fake profiles to extract information, or create chaos by spreading false alerts en masse (as in the Pentagon case above). The human-machine distinction becomes a security issue: we already see it with the multiplication of CAPTCHAs and other verification systems to ensure a user is really human. Automated social engineering must also be taken into account: AIs conversing in real time can deceive and manipulate employees or customers. In this context, training teams to spot suspicious interactions and strengthening authentication protocols is essential.
In short, whatever your sector in digital, the “dead internet” phenomenon (in the sense of a web saturated with fake content) forces you to adapt your practices. Whether to reach an audience, analyze online behavior, train a model, or secure a system, you have to work with an ecosystem where the artificial share has become far from negligible.
▸ Conclusion: preserving the integrity of the web, a collective responsibility
The Dead Internet Theory, taken literally, remains a conspiracy theory with its share of exaggerations. No, the Internet is not entirely “dead”: billions of human beings keep creating, sharing, and exchanging on it every day. However, the trends it points to should not be ignored. The current state of the web is the result of our technological and economic choices: advertising pushed at all costs, chasing audience growth by any means, deploying AI without guardrails… It would be too easy to declare the death of the Internet an inevitable fate.
On the contrary, it is a wake-up call: it is up to us, as digital professionals but also as ordinary users, to preserve the integrity of the web. That runs through several avenues of thought and action. First, encouraging platforms toward more transparency (about sponsored content, about the use of recommendation algorithms, about the presence of bots), and even regulation to limit abuse. Then, developing and supporting tools for detecting fake content and automated accounts, in order to give quality authentic content its visibility back. It is also the responsibility of creators and communicators not to give in to the ease of click farms and full automation: putting the human back in the loop, valuing truthfulness, originality, and reliability, are editorial and ethical choices that pay off in the long run.
As a digital community, we must think about the Internet we want for tomorrow. Do we really want an artificial desert where “the machines talk to each other” and the user is a spectator of an illusion? If that picture is frightening, it means there is still hope: the hope of proving the prophecy wrong. By becoming aware of the current drift and acting collectively (technicians, decision-makers, content creators, and users), we can work toward a human-scale web, where artificial intelligence remains a tool in the service of humans and not an uncontrolled flood. The ball is in our court: it is up to us to keep the web well and truly alive, diverse, and authentic, for the generations to come.
▸ Further reading
Ouest-France. Une théorie du complot affirme qu'Internet est « mort » depuis 2016. Ouest-France, September 6, 2021.
The Atlantic. Foley, Jason. Maybe You Missed It, but the Internet “Died” Five Years Ago. The Atlantic, August 13, 2021.
Imperva. Bad Bot Report 2021: The Pandemic of the Internet. Imperva, April 2021.
Forbes. Why Native Advertising Is The Future Of Online Marketing. Forbes, January 23, 2021.
The Guardian. Paul, Kari. How Many Twitter Users Are Bots? Elon Musk Claims Far More Than the Company Admits. The Guardian, May 17, 2022.
AP News. AI-generated image of explosion near Pentagon may have been created by Russian accounts, analysts say. AP News, May 23, 2023.
Epsilon Magazine. Les bots, bientôt maîtres de Wikipédia ? Epsilon, October 4, 2022.
France Inter. La théorie de l'Internet mort : pourquoi certains pensent que le Web n'existe plus vraiment. France Inter, September 9, 2021.
Wikipedia. Dead Internet Theory. Last accessed: July 2025.
Does this resonate with your team?
Let's talk about how Atypical Consulting can help you move forward.
Contact me