ai
Share of AI-Generated Online Articles Stabilizes Near 50 Percent
Graphite.io analysis of more than 55,000 sampled web pages shows AI-generated content surged after ChatGPT's November 2022 launch but has held steady near half of new articles for more than a year. The study used three AI detectors on English-language articles and listicles from Common Crawl. Researchers warn of potential feedback loops as models train on machine-generated text.
May 15, 5:00 AM·1m read1 source
app.buzzsumo.comAudio version
Tap play to generate a narrated version.
Developing·Limited corroboration so far. This page will refresh as more sources emerge.
Io. Graphite randomly sampled 55,400 English-language URLs from Common Crawl, a large public archive of the web often used in AI research and training datasets. The sampled pages were at least 100 words long, had publish dates between January 2020 and March 2026 and were classified as articles or listicles.
The analysis is based on an average of three AI-detector tools: Pangram, GPTZero and Copyleaks. Graphite classifies an article as primarily AI-generated only when most of its text is detected as AI-written or AI-assisted. 9 percent within one year of ChatGPT's release.
ChatGPT launched in November 2022. But since early 2025, the share has hovered at around 50 percent. The share of primarily AI-generated online news articles, blog posts and listicles has held near 50 percent for more than a year.
The plateau indicates that rapid growth in machine-generated content has stalled after the initial surge that followed ChatGPT's debut. Axios reported that counting AI-generated writing is still messy because many articles are no longer written purely by humans or AI. A human may use AI for outlining, drafting, rewriting or editing.
This makes the line between "AI" and "human" fuzzier than simple detection suggests. Dan Klein, a UC Berkeley professor and AI model CTO, said these models are smart because of all the information we put on the web that was created without these models.
Researchers who have studied the spread of AI-written articles warn that once models start training on that content, the internet could become a massive feedback loop of low-quality, machine-generated content.
