One site finds AI crawlers are 24% of bot traffic, exposing a measurement gap
A live server-log case study shows why “AI traffic” is too blunt a category for access decisions.
Mojo Dojo’s live index counted 14,523 AI-crawler requests among 61,664 bot hits over 30 days, including 1,041 pages fetched by assistants in response to user questions. It separates training, AI-search and user-fetch traffic and checks published IP ranges or reverse DNS where possible, although the figures cover one website and IP-based verification began only on October 7. Publishers need this request-level distinction to decide which machine access represents training, discovery or potential audience demand before setting access and licensing terms.