The digital ecosystem is mutating rapidly. My analysis of fresh data shows: roughly every tenth web page in open access bears clear traces of generative AI. However, this is just the tip of the iceberg—if you filter out "prehistoric" content created before the era of large language models, the picture becomes far more alarming for traditional authors.
Scale of the phenomenon: from 10% to 34%
I examined a sample of 10,000 random English-language pages collected in July 2026. The methodology was based on machine learning that identifies characteristic linguistic patterns inherent to generative models. Among all pages, the share of "synthetics" was about 10%. But that's just the average temperature across the board.
The key insight lies elsewhere. If we consider only materials published after ChatGPT's debut in November 2022, signs of AI authorship are found in more than a third of pages. This means we are witnessing not just a trend, but a fundamental shift in internet content production, accelerating with the release of each new model—from Claude to Gemini.
Geography of synthetics: commerce leads
The distribution of AI content across domain zones is extremely uneven, and that's expected. In the .com zone, signs of generation are visible on every tenth page, which is twice the rate of .org (4.6%). On academic resources (.edu) and government portals (.gov), the share drops to a modest 1%.
This disparity is explained by the economics of content. Commercial platforms actively automate routine tasks: product descriptions, SEO texts, reference articles, and news briefs. These formats are the most susceptible to automation, whereas unique research and official documents require human involvement.
Methodological nuances
It's important to emphasize: this assessment is not an exact count, but rather a probabilistic model. Researchers did not analyze edit histories or survey authors. The method captures linguistic markers that statistically correlate with AI generation. Additionally, the category includes both fully generated texts and materials that have undergone substantial AI editing. Light proofreading does not fall into this category.
My comment: These numbers are a wake-up call for the entire industry. We are moving toward a world where search engines and readers will drown in a flood of template content. The irony is that high-quality human texts will become scarce and, consequently, gain premium value. The question is not whether AI will displace authors, but who can offer unique expertise that algorithms cannot replicate.