Running coverage · Updated July 8, 2026
The State of AI Content and Brand
The running record of what AI-generated content is doing to differentiation: how much of the web machines now write, where the clicks went, and which moats survive. Updated as data lands.
The short version
- Machines now write more of the internet's new articles than people do. Graphite's crawl analysis put AI-generated articles at 51.7% of everything published online by May 2025, and Ahrefs found AI text on 74.2% of 900,000 newly published pages.
- Volume stopped paying: Google's AI summaries cut the click rate roughly in half (8% vs 15% per Pew's tracking of real searches), so more competent content chases fewer visits.
- Ranking still favors humans. Even with AI writing half of new pages, Graphite found 86% of the pages that actually rank in Google are human-written.
- The surviving advantages are the ones a model can't synthesize from public text: a recognizable brand, an owned audience, and a publication that can prove what it prints.
When the cost of a competent article fell to near zero, the value of a competent article followed it down. This page is our running record of that collapse and what survives it: how much of the web machines now write, where the clicks and the trust went, and which marketing assets still can’t be generated.
What the evidence says so far
The supply side flipped first. By Graphite’s count, machines were writing more of the internet’s new articles than people by late 2024, and Ahrefs found AI text on three of every four newly published pages. The result isn’t chaos; it’s sameness. Feed the same models the same prompts and every brand in a category converges on one voice. Our full read: AI made content almost free to produce, and differentiation almost impossible.
The demand side confirms the squeeze. Google now answers many queries in place, and Pew’s tracking of real user behavior shows the click rate falling by roughly half when an AI summary appears. Content teams are producing more and getting less, which is why brand and an owned audience are the last moats a commodity content engine can’t manufacture. The distribution side of this shift, who gets cited when search becomes an answer, is covered in our AI marketing stack coverage.
The owned-media boom runs straight into the same wall. Standing up a company publication is now a budget line rather than a team, which means the flood of fluent, sourceless brand content is just getting started. The publications that survive it will be the ones that can prove what they print: trust is the product, the writing is table stakes. We run this publication on those rules, on the record, because that’s the bet.
The pattern to watch
Three numbers tell you where this goes: the share of new content machines write (rising), the click rate on the content everyone is producing (falling), and the share of ranking pages that are still human-written (86%, per Graphite). As long as the third number holds, the work a model can’t fake, original data, named authors, earned trust, keeps compounding while volume keeps deflating.
We update this page as new data lands: crawl studies, click-rate research, brand-spend surveys, and on-the-record accounts from the people rebuilding content teams around this shift. If you run content or brand at a B2B company and have numbers from inside the change, get quoted.
Questions people ask
How much web content is AI-generated in 2026?
Two large studies set the range. Graphite's classification of Common Crawl pages found AI-written articles crossed 50% of new online articles in late 2024 and hit 51.7% by May 2025. Ahrefs analyzed 900,000 newly published pages and found 74.2% carried at least some AI-generated text. Barely a quarter of new pages are written by a human alone.
Does AI-generated content still work for marketing?
As a floor, yes; as a differentiator, no. When every team runs the same content engine, output converges on one voice, and Google's AI summaries halve the clicks that content used to earn. The pages that still win carry things a model can't assemble from everyone else's text: original data, a real point of view, a named author.
What is the marketing moat when AI commoditizes content?
Brand and owned audience. A brand is memory sitting in the buyer's head before any search happens, so it skips the auction. An owned audience (email list, subscriber base, community) is distribution no algorithm change can revoke. For publishers, add verifiable trust: sourcing you can prove beats volume you can generate.
Get Net Effect.
The net effect of AI on your marketing: the stories that matter, twice a week, in five minutes.
