Home / Business and Politics / The Dead Internet Theory: Digital Reality Under AI Control

The Dead Internet Theory: Digital Reality Under AI Control

mrtvi internet, društvene mreže
mrtvi internet, društvene mreže / Image by: foto Shutterstock

Sam Altman, the CEO of OpenAI, recently sparked a flurry of reactions on X when he acknowledged the possible credibility of the so-called ‘dead internet theory’. This theory, which sounds like it came from some dark corner of the internet, has often been framed as a conspiracy. The theory suggests that a large part of today’s internet, especially social media, is actually dominated by artificial (non-human) activities, such as AI-generated content, bots, and corporate interests, leading to a reduction in authentic human interaction. The content is produced by computers, LLMs, and bots, and with the emergence of AI platforms like Altman’s ChatGPT, this process has accelerated further. Sam Altman stated that he had not previously taken this idea, i.e., theory, seriously, but is now aware of numerous profiles online run by large language models.

Truth or Myth

The term ‘dead internet’ unfortunately carries a truth that many today feel acutely. If you are part of the generations that grew up with the internet, who listened to the sound of modems connecting to phone lines, you likely sometimes look back nostalgically at those ‘early’ days of the global internet. The internet of the 1990s and early 2000s was messy, eccentric, and spontaneously creative. Forums, fan pages, chats, and blogs flourished in a digital ‘wild commons’ that was quite chaotic, but very human and above all experimental. However, just as radio and television once succumbed to advertisers and regulators, the free internet could not last forever. The vibrant web has been replaced by the web we have today, a place where attention is currency, and algorithms determine what you like, what is popular, and what is sought after.

The internet no longer operates vibrantly; yes, it functions and is more filled than ever with content and activity, but that energy and humanity have slowly but surely faded. What mostly remains is a system driven by machines optimized for profit.

The Economy of Decay

Profit has destroyed the internet. We killed it because we wanted it to earn money. As we said, attention is currency, and advertisers pay for clicks, then quality, if at all. ‘SEO farms’ are websites that create a large number of low-quality articles and links to manipulate search engine results, targeting solely Google’s algorithms, while social networks optimize content that encourages compulsive scrolling, not to mention all the mental and other health issues arising from social media. Now this modus operandi is accelerated by artificial intelligence, and AI models are churning out ‘tons’ of bad and false content.

The authors of the paper ‘The Dead Internet Theory: A Survey on Artificial Interactions and the Future of Social Media’ state that up to 60 percent of web traffic is generated by bots. Studies cited in the paper estimate that nearly two-thirds of internet traffic comes from automated systems. These bots like, share, comment, and create content to mimic human behavior. The authors state.

Where Has the Human Gone

They write about social media favoring metrics over people, which any older Facebook user could notice when, at one point, almost all their friends disappeared from their feed, and sponsored posts, strange pages, and mostly poor content took precedence.

The authors in the paper note that platforms like Facebook, TikTok, and X are increasingly investing in algorithms that maximize time spent on the platform, which directly increases advertising revenue. This leads to sensationalism, homogenization, and a loss of diversity. Also, instead of quality and original content, algorithms encourage the generation of large amounts of repetitive, superficial, and often clickbait material whose goal is not to inform or connect users but to keep the user engaged for as long as possible and, if possible, provoke some reaction, preferably negative. The research also highlights a warning about ‘astroturfing’, i.e., situations where corporations or governments create fake profiles and campaigns to influence public opinion, masquerading as authentic users.

Data on the state of the internet shows that the digital world is rapidly changing under the influence of automation. According to the Imperva Bad Bot Report, an annual report published by Imperva, a cybersecurity company, for the first time in the last decade, automated traffic has surpassed human traffic, accounting for 51 percent of total web traffic in 2024. Of that, 37 percent was recorded as malicious traffic, while only 14 percent was marked as benign bots. Similar estimates are provided by SOAX, an international company that provides proxy and data services for monitoring internet traffic, which states that in 2023, bot traffic accounted for nearly 50 percent of total internet activity, distributed across 32 percent malicious and 17.6 percent benign bots. According to data from Akamai Technologies, a global leader in cloud services and network infrastructure, bots account for 42 percent of web traffic, with 65 percent of activity being malicious.

AI Will Kill Us

Content produced by AI and bots is also becoming increasingly dominant. Experts predict that by 2026, as much as 90 percent of internet content could be synthetically generated, while an academic study by Spennemann from 2025 shows that at least 30 percent of text on active websites has already been created by AI, with the actual share possibly reaching 40 percent. Even on social media, an analysis of activity during the 2024 U.S. presidential elections showed that about 12 percent of images and 1.4 percent of text were AI-generated.

Changes are also occurring in how the internet indexes content. New retrieval bots used by OpenAI, Anthropic, and others have recorded a 49 percent increase in traffic to websites between the end of 2024 and the beginning of 2025, searching the web to generate summaries for users instead of classic Google results. In the music industry, although AI-generated records account for only 0.5 percent of all streams on the Deezer platform, as much as 70 percent of such streams represent fraud, generated using bots for covertly extracting royalties.

All of this also affects user trust. According to a New York Post survey, only 41 percent of Americans believe that online content is true, while 23 percent believe it is completely false, and 36 percent approach it with caution. As many as 78 percent of respondents admit they have difficulty distinguishing between human and AI-generated content, while nearly half of them, 48 percent, believe that posts on social media are driven by artificial intelligence.

Meme or Propaganda

At first glance, the motivation behind the mass spread of AI and bot content may seem simple: greater interaction generates advertising revenue. However, when you scratch the surface, a more serious threat emerges: political propaganda and manipulation of public opinion. A 2018 study analyzing 14 million tweets found that bots systematically spread articles from unreliable sources, often legitimizing misinformation. Similar patterns were observed after mass shootings in the U.S., when bots on the X platform took over the discussion, amplifying or distorting narratives related to the events.

The Economy of Automation

YouTube and TikTok are increasingly flooded with AI-generated videos, as we have already written in Lider. From travel guides and home recipes to deepfake vlogs where virtual influencers run channels with millions of views, to AI ads between videos. (Some AI Tai Chi guy drove me crazy). Such content is created at minimal cost, automated voices, generated animations, and fake comments maintain the illusion of community. For example, a series of channels on YouTube publishes dozens of videos daily, often recycled or entirely created by artificial intelligence, attracting thousands of views and a share of the advertising pie.

One of the recently highlighted examples in global media of this new generation of content is the channel Boring History for Sleep. This channel publishes hours-long videos dealing with historical topics, often with titles like ‘How Medieval Peasants Survived the Coldest Nights’, or ‘Why You Wouldn’t Last a Day in Medieval Times’. Although they may seem like educational content at first glance, they are AI and stories that are not necessarily related to accurate historical events. (AI slop) Despite this, some of these videos have attracted millions of views, raising concerns among historians and educators who are engaged in creating authentic and investigative content. And there are countless such channels. Additionally, the flood of children’s channels with AI cartoons that, in most cases, make no sense is slowly becoming a problem for parents.

Dead or Dying

It should be clear: the ‘dead internet’ is not a literal claim that everything we do online is false. However, it becomes a useful metaphor; the internet is increasingly less a space ‘by people, for people’. The content we consume is increasingly shaped by algorithms, bots, and generative networks, while real users lose dominance. The danger is that such artificial patterns can be used to manipulate opinions, political decisions, and social perceptions, and the average user struggles to recognize where human influence ends and automation begins. As researcher Vlada Rozova warns: ‘Any interaction, trend, or ‘general sentiment’ online can be synthetically designed to change the way we perceive the world.’ The internet as we knew it, an open space for human creativity, discussion, and knowledge sharing, is today under strong influence of automation and increasing control. Until perhaps some new decentralized nostalgic platform. Ultimately, the internet is not ‘dead’, but it is seriously ill and a warning of the direction we are heading if we do not change the way we build and experience digital spaces.

Tagged: