AI-Generated Articles Constitute 52% of New Online Content, Graphite Study Reveals

A recent study by SEO firm Graphite indicates that 52% of new articles published online are now primarily generated by artificial intelligence. This surge in AI-produced content has led some to dub the phenomenon a "Slopocalypse," as noted by social media user Crémieux. The report, which analyzed web content through May 2025, highlights a significant shift in the digital publishing landscape, though its impact on search engine visibility remains nuanced.

Graphite's analysis involved a random sample of 65,000 English-language articles published between January 2020 and May 2025, sourced from the Common Crawl database. Utilizing an AI detection tool called Surfer, articles were classified as AI-generated if 50% or more of their content was identified as machine-written. This methodology aimed to provide a quantitative measure of AI's growing footprint in online text.

The proliferation of AI-generated articles dramatically accelerated following the public launch of ChatGPT in November 2022. Within a year, AI-authored content accounted for approximately 39% of new publications, eventually surpassing human-written articles in November 2024. However, the growth rate has since plateaued, hovering around the 50-52% mark, suggesting a stabilization in the influx of purely AI-generated text.

Despite the high volume of AI-generated content, its performance in search engine rankings appears limited. A separate Graphite study found that only about 14% of articles appearing in Google Search results are AI-generated, with 86% still attributed to human authors. This suggests that search engines are effectively filtering out lower-quality AI content, preventing it from dominating top results.

The findings raise questions about content authenticity and the evolving relationship between human and artificial intelligence in writing. While AI offers a cost-effective alternative for content creation, especially for general-interest topics, the market's reception and search engine algorithms continue to favor human expertise. The study also acknowledges the increasing prevalence of AI-assisted human-edited articles, blurring the lines between machine and human authorship.