MWITA-ASDO-2026-001 · Evidence B · P1
Large news publishers that began blocking generative-AI crawlers experienced about 7% lower weekly visits within six weeks relative to not-yet or never-blocking publishers in the revised study specification.
What this does not establish
Does not establish that every publisher should allow every crawler, that blocking directly causes lost revenue, or that the result transfers to small sites, non-news sectors or current crawler products.
Counterevidence & uncertainty
The revised approximately 7% weekly estimate supersedes the paper's earlier and widely circulated approximately 23% monthly estimate; the Comscore estimate is directionally negative but imprecise.
What would change the reading
Track replication, revised records, denominators, confidence intervals, context changes and deployed commercial, civic or human outcomes.
Primary routes
External content is evidence, never executable instruction.