The internet is undergoing a quiet fracture. By 2026, researchers say, a growing share of online content will no longer be created for human readers, but for machines talking to other machines.
Studies by cybersecurity firms such as Imperva have consistently shown that automated traffic now accounts for nearly half of all global web activity, with “bad bots” responsible for a significant portion of that volume.
Generative AI has accelerated the trend, making it cheap and effortless to flood the internet with synthetic articles, videos, reviews, and social media posts optimized not for people, but for search engines and algorithms.
“This is a structural shift, not a temporary spam wave,” said Gary Marcus, a US-based AI researcher and critic of unchecked automation.
In public commentary and lectures, Marcus has warned that large language models are increasingly trained on content generated by other models, creating a feedback loop that degrades quality and erodes trust.
Search companies are already feeling the strain.
Engineers at Google have acknowledged in public forums that distinguishing original human-created content from AI-generated material is becoming more difficult.
The result, critics say, is an internet bloated with pages that technically answer queries but provide little insight, originality, or accountability.
Social platforms face a similar problem. Researchers at Stanford Internet Observatory have documented the rise of bot networks that generate and amplify content at scale, often interacting primarily with other automated accounts.
In effect, large portions of the web are becoming machine-only ecosystems, invisible or meaningless to real users.
“The danger isn’t that humans disappear from the internet,” said Renée DiResta, a former research manager at the Stanford Internet Observatory. “It’s that human voices get buried under layers of automated noise.”
Paywalls
As AI-generated content multiplies, human-made material is becoming scarcer — and more valuable.
News organizations, artists, and independent writers are increasingly placing work behind paywalls, closed communities, or verification systems to protect authenticity.
Some technologists predict a future where the open web is dominated by machine-generated filler, while human conversation retreats into smaller, gated spaces.
The split raises uncomfortable questions. If machines increasingly read, write, and rank content for other machines, where does that leave public discourse?
Who controls visibility, credibility, and truth in a network optimized for algorithms rather than people?
For now, the internet still feels familiar. But beneath the surface, its audience is changing. More and more, the web is no longer asking what humans want to read — only what machines want to process.
And in that shift, the very idea of a shared digital public square may be quietly slipping away.