Artificial intelligence is no longer simply helping people navigate the internet. It is increasingly creating the content they find there.

More than one-third of the English-language web pages published since ChatGPT launched in November 2022 show significant signs of AI authorship, according to a new study from Pew Research. The findings add to growing evidence that generative AI is rapidly changing how online content is produced, distributed and consumed.

AI-generated content is becoming part of the web’s infrastructure

For its analysis, Pew used the Common Crawl web archive to examine nearly half a million English-language pages collected over approximately five years. The dataset covered content published both before and after the public release of ChatGPT.

Researchers then used technology developed by Open Pangram to identify pages that were likely generated or substantially edited using artificial intelligence.

In a random sample of 10,000 pages collected in July 2026, approximately 10% displayed significant indicators of AI authorship. However, that sample also included older pages published before widely accessible generative AI tools existed.

When researchers excluded those pages and focused exclusively on content published after ChatGPT’s release, the proportion increased to 35%.

The result suggests that AI-assisted publishing has moved beyond isolated experimentation and become a significant component of the modern web.

Commercial websites appear to use AI most heavily

The study also identified considerable differences between types of domains.

Pages hosted on .com domains displayed signs of AI authorship at roughly ten times the rate observed on .edu and .gov websites. Only around 1% of the analyzed content on educational and government domains appeared to be AI-generated, while the rate for .org pages was approximately 4.6%.

This gap may reflect the different pressures facing commercial publishers. Businesses, media organizations and marketing teams are expected to produce increasing volumes of content across websites, social platforms, newsletters and search channels. Generative AI offers a fast and relatively inexpensive way to scale that output.

The challenge is that greater publishing capacity does not automatically produce better information.

A web increasingly created and consumed by machines

The findings arrive shortly after Cloudflare reported that automated bot traffic had surpassed human web traffic. Taken together, the two developments point toward an emerging online ecosystem in which machines generate a growing share of the content while other machines crawl, summarize, rank and reuse it.

AI systems may therefore be trained on, or retrieve information from, material previously created by other AI tools. Without strong quality controls, this cycle can amplify factual errors, repetitive language and low-value content across the internet.

For companies, the problem extends beyond readability. AI-generated content that lacks expert oversight can damage search visibility, weaken brand credibility and introduce inaccurate or unsupported claims into customer communications.

AI detection remains an imperfect science

Pew acknowledges that the methodology has limitations. AI-detection systems can incorrectly classify human-written content as machine-generated, particularly when the writing is structured, polished or formulaic.

The research should therefore be interpreted as a directional indicator rather than a precise measurement of every AI-written page online.

The study also found that certain stylistic patterns frequently associated with generative AI have become more common. These include em dashes, Oxford commas and constructions such as “it’s not X, it’s Y.” None of these features proves that a text was created by AI, but their increased frequency illustrates how generative tools may be influencing the broader style of online writing.

What this means for businesses

The strategic question is no longer whether organizations should use AI to create content. Many already do. The more important issue is how they can use it without losing accuracy, originality and trust.

AI can accelerate research, structure information, generate initial drafts and help teams adapt content for different audiences. It should not replace subject-matter expertise, editorial judgment or fact-checking.

As the volume of machine-assisted content increases, distinctive and credible communication becomes more valuable. Organizations that combine AI efficiency with human knowledge, clear perspectives and rigorous review will have a stronger chance of standing out in an increasingly automated digital environment.

The future of online content will almost certainly involve AI. The competitive advantage, however, will belong to businesses that use the technology deliberately rather than simply using it to publish more.

Source