.png){kind=logo}
AI tools are making large-scale content production faster, cheaper, and increasingly accessible.
Research shows AI-assisted webpages are becoming more common across the internet.
Experts continue debating AI content's effects on quality, originality, and information diversity.
The internet is filling up with content that can be produced in seconds. Articles, images, videos, and entire webpages can now be created or substantially edited using artificial intelligence, making online publishing faster and cheaper. However, the rapid growth of AI-generated material is also raising questions about quality, originality, and the character of the web itself.
The term ‘AI slop’ has emerged to describe large amounts of low-quality or low-effort AI-generated content. It generally refers to material produced with little meaningful human involvement and little value for the audience. The concern is not that every piece of AI-assisted content is poor, but that generative AI has made mass production of mediocre content significantly easier.
A recent Pew Research Center analysis provides a clearer picture of how widespread AI-assisted content has become. Researchers examined 490,000 English-language webpages from the Common Crawl web archive, covering 49 crawls between January 2021 and July 2026.
In a random sample of 10,000 webpages collected in July 2026, 10% showed significant signs of AI authorship. The figure was considerably higher among pages published after ChatGPT was released.
For webpages published after November 2022, more than one-third showed signs of AI authorship in the July 2026 snapshot. Pew noted that AI detection tools are not perfect and can misclassify individual pages, but patterns across large collections can provide a broader picture.
The increase is also visible across domain types. In 2026, around one in 10 .com pages showed signs of AI authorship, compared with 4.6% of .org pages and about 1% of .edu and .gov pages.
AI slop is not simply any content created with artificial intelligence. AI can assist with research, editing, translation, design, and other tasks while a human remains responsible for the final work.
The concern arises when AI is used primarily to produce large volumes of content quickly, often without original reporting, meaningful editing, or independent verification.
That can include generic articles created for search traffic, repetitive social media posts, automatically generated images and videos, or websites assembled around large quantities of AI-written material.
As the cost of producing such content falls, publishers and creators can generate significantly more material without increasing their human workforce at the same rate.
Also Read: LinkedIn is Now Social Media's Top Hub for AI Slop: New Study
Research from Imperial College London, the Internet Archive, and Stanford University examined the broader effect of AI-generated and -assisted text on online content.
The study found that by the first half of 2025, around 35% of newly published websites were classified as AI-generated or AI-assisted. The researchers also examined whether the growing use of AI was changing the diversity of online information.
One finding concerned semantic diversity. AI-generated or AI-assisted websites had nearly 33% higher semantic similarity than human-made websites. In simple terms, AI-produced material tended to be more similar in meaning and ideas.
The research also found that the average positive sentiment score of AI-generated or AI-assisted websites was 107% higher than that of non-AI websites. Researchers linked the finding to the tendency of current large language models to produce agreeable and optimistic responses.
However, the study did not confirm every concern surrounding AI content. Researchers found no statistically significant evidence that increased AI-generated text reduced factual accuracy or stylistic diversity.
Pew’s analysis found that certain words, phrases, and punctuation associated with AI-generated writing are becoming more common online.
Compared with 2023, em dashes appear about twice as frequently, while the use of Oxford commas has increased by 63%. Certain words frequently associated with AI writing, including ‘delve,’ ‘interplay,’ and ‘testament,’ have more than doubled in usage.
Pew also found that the use of a writing structure known as ‘negative parallelism’, comparisons built around constructions such as ‘it’s not just X, it’s Y’, has nearly tripled, although it remains relatively uncommon overall.
These characteristics do not prove that AI wrote an individual article. Humans also use the same words and punctuation. The patterns become more meaningful when examined across hundreds of thousands of webpages.
Also Read: TikTok Beats YouTube in AI Slop Race, Reveals Survey Report
The increase in AI slop does not mean the internet is becoming entirely machine-generated or that AI-assisted content has no value.
AI can reduce production time, help small teams create material, and make creative tools accessible to more people. Human oversight can also substantially change the quality of AI-assisted work.
The larger concern is the incentive to produce content simply because it is cheap to make. When websites can publish thousands of pages with limited human involvement, the internet can become harder to navigate for people looking for original reporting, specialist knowledge or genuinely useful information.
AI has changed the economics of online publishing. Content that once required hours of writing, editing or production can now be generated in minutes. This creates opportunities for creators and businesses, but it also makes low-value content easier to produce at scale.
The challenge for the internet is therefore not simply to identify whether AI made something. It is to maintain the value of original reporting, human judgment, verification, and distinctive ideas as synthetic content becomes a larger part of what people encounter online.
What is AI slop?
AI slop refers to large volumes of low-quality, low-effort content generated using artificial intelligence with limited human involvement or oversight.
Why is AI slop increasing online?
Generative AI makes producing articles, images, videos and webpages faster and cheaper, encouraging large-scale automated content creation.
Does all AI-generated content qualify as AI slop?
No. AI-assisted content can be valuable when humans provide original research, editing, verification, creativity and meaningful oversight.
How can AI slop affect the internet?
It can make useful information harder to find, increase repetitive content and potentially reduce the diversity of online material.
Can AI slop be identified easily?
Not always. AI detection tools can make mistakes, while human writers may also produce patterns commonly associated with AI-generated content.
