
The Dead Internet Theory claims that bots and AI-generated content now dominate the web, pushing real people to the margins. The broad trend checks out. Automated bots made up 53% of all web trafficThe number of visitors to a website, often used as a measure of an affiliate's reach and influence. in 2025, the second straight year machines outnumbered humans online. For brands, authentic human content has become the clearest way to stand out and earn trust.
The Dead Internet Theory is the idea that most online activity since roughly 2016 has come from bots and automated content, not real people. Earlier versions circulated on imageboards, but the theory’s best-documented public formulation appeared in a 2021 post on Agora Road’s Macintosh Cafe. It reached a wide audience after a 2021 Atlantic article argued the internet had quietly “died.”
The theory has two versions. The strong version is a full conspiracy. It claims that much of the modern web has been replaced by bots and synthetic activity, sometimes as part of a coordinated effort to manipulate public opinion. There is no credible evidence for that broader conspiracy. Few researchers take that seriously. The weak version is harder to wave away. It says automated accounts and AI text increasingly shape what people read, watch, and believe. That part is measurable.
Synthetic content means any media generated or altered by AI, from text and images to video and cloned voices. You have likely seen it without naming it. In 2024, FacebookA social networking site where users can post comments, share photographs, and links to news or othe... feeds filled with surreal AI images like “Shrimp Jesus,” which pulled tens of thousands of likes from automated accounts. The odd part was not just the picture. It was the surrounding ecosystem of AI-generated spamUnsolicited and irrelevant emails sent to a large number of recipients., engagementThe interactions that users have with a brand’s content on social media. farming, automated accounts, and users who could not always tell whether they were interacting with people or machines.
Less than half. For the second year running, automated traffic outran human activity, and the gap keeps widening.
Traffic type | 2023 | 2024 | 2025 |
|---|---|---|---|
Human traffic | 50.4% | 49% | 47% |
Bad bots (scraping, fraud, attacks) | 32% | 37% | 40% |
Good bots (search and AI crawlers, monitors) | 17.6% | 14% | 13% |
Source: Imperva / Thales Bad Bot Reports, 2024 to 2026
Bad bots alone now make up 40% of traffic, a seventh straight year of growth. Cheap, capable AI models let almost anyone run convincing automationUsing software to send emails automatically based on predefined triggers and schedules. at scale.
The content layer tells the same story. An Ahrefs study of 900,000 new web pages found that about 74% of pages published in April 2025 carried AI-generated text. Only a quarter were purely human. The lesson for marketers is blunt. Publishing volume used to signal effort. It signals nothing now, because anyone can generate hundreds of articles in an afternoon.
Scarcity creates value, and genuine human work is now the scarce input. As feeds fill with synthetic media, audiences reward the brands that still sound like people.
The numbers back this up. Fifty-five percent of social media users say they are more likely to trust brands committed to publishing human-created content, rising to roughly two-thirds among Gen Z and Millennials, according to Sprout Social. More than half, 52%, worry about brands using AI without disclosing it. The cost of getting it wrong is real. Half of Gen Z will block or unfollow an account that posts low-effort AI content, and roughly six in ten consumers say they are less likely to engage with brand content in today's AI-heavy climate.
Here is what makes content read as unmistakably human:
None of this means dropping AI. The approach that wins is hybrid. Use AI to draft, research, and speed up production, then put human experts in charge of judgment, accuracy, and emotion. Done right, that is how you create AI-assisted content that still passes E-E-A-T. Scott Morris, CMO of Sprout Social, calls transparency “the cornerstone of trust” when audiences want to know how content was made.
The same logic applies to AI search. Google and the large language models judge trust and authority using signals that machine-spun text rarely earns: real experience, citations, and original insight.
The internet is not literally dead, but it is more crowded with machines than ever. That shift is an opening. While competitors flood the web with generic output, brands that publish genuine, expert-led work will own the trust automation cannot manufacture.
If you want help scaling content production with AI without losing quality that stands out in an AI-saturated feed, the team at Bliss Drive can map the strategy with you.
The conspiracy version is not proven, and most researchers reject it. The core trend is real and measurable. Bots made up 53% of web traffic in 2025, and most new web pages now contain AI-generated text. Human activity has not vanished. It is simply the minority of raw online traffic for the first time in the web's history.
Bots accounted for 53% of all web traffic in 2025, up from 51% in 2024, according to Imperva's 2026 Bad Bot Report. Of that, 40% came from bad bots used for scraping, fraud, and attacks. About 13% were good bots, like search and AI crawlers. Human traffic fell to 47%.
Not on its own. Google rewards helpful, people-first content no matter how it is produced. The real problem is thin, mass-produced AI text that adds nothing new. That content struggles to rank and rarely earns citations. The data on AI content and Google shows a consistent pattern: originality and expertise win.
Show your work. Use named authors with verifiable credentials, include first-hand experience and original data, link to reputable sources, and be honest about trade-offs. When you use AI, say so. These signals are hard for automated content to copy, which is why both audiences and search engines treat them as proof of trust.
