Pew Research Center scanned approximately 490,000 web pages from the Common Crawl database between January 2021 and July 2026 and found that 1 in 10 English-language pages contained AI-generated traces. This machine dominance rate jumps to 35% specifically for pages published after the launch of ChatGPT in November 2022, according to test results from the Open Pangram detector tool.
This surge in machine-generated text is most prominent in the commercial corners of the internet. Pages with the .com domain extension show an AI-writing ratio 10 times higher than academic (.edu) or government (.gov) websites, which both remain at around 1%. This ratio is also twice as high compared to organizational sites (.org), which recorded an AI involvement rate of 4.6%. For the .com domain group alone, the trace of AI writing climbed from around 1% in January 2021 to 9.35% in early 2026 testing.
What Are the Machine’s Tell-Tale Signs?
The Open Pangram detector tool used by Pew exposes machine habits that have become increasingly common since 2023. Specific punctuation like em dashes (long hyphens) now appears twice as often in texts, while Oxford comma usage has increased by 63%. English vocabulary words that were once rarely used, such as “delve”, “interplay”, and “testament”, are also appearing twice as often in various internet articles.
In terms of writing structure, the presence of negative parallelism, or contrasting sentence patterns with the structure “not only X, but Y”, has skyrocketed nearly threefold since 2023. These repetitive patterns are not limited to personal blogs or low-tier websites. Open Pangram reports that signs of AI presence are attached to around 9% of United States daily newspaper articles this year, where the traces have also slipped into the opinion pages of established print media like the New York Times.
Direct Impact on the Crypto Ecosystem
This flood of unfiltered machine content is beginning to burden organic search systems on the internet. This saturated situation prompted Merriam-Webster lexicographers to designate “slop” - a term for digital junk content - as the word of the year last December. For the crypto industry and the Web3 ecosystem, this shift in information presents real problems in daily operations.
The pile of low-quality web pages flooding the internet is damaging the SEO landscape for crypto projects. Investors are finding it increasingly difficult to filter out valid information due to the high volume of generic articles. As a future mitigation step, AI companies like Anthropic are now working to embed text fingerprint features at the base level of their models to make text identification easier, so that readers will no longer have to guess the origin of the readings on their screens.
As reported by Decrypt.
Read also: What Is DeFi (Decentralized Finance)?
Disclaimer: This article is for informational and educational purposes only, not financial advice. Cryptocurrency assets are highly volatile and carry significant risk. Always do your own research (DYOR) and never invest more than you can afford to lose.
