Seven in ten Canadian news sites block AI crawlers, banks block none
Seven in ten Canadian news websites now instruct AI crawlers to stay out of their content, while not a single bank, insurer or fintech in the country does the same, according to a study published on 20 October 2026 by Chateau Mojo, a Quebec-based SEO and AI search consultancy. The findings expose a sharp divergence in how different industries are approaching AI data access, with implications for what AI-powered tools can and cannot surface about Canada's economy, culture and public life.
The study examined robots.txt files across 388 of Canada's best-known websites on 17 October 2026, spanning news, banking, government, retail, telecom and education, among other sectors. Of the 372 files that could be read, 21% block at least one AI crawler from the whole domain. That figure balloons to 70% when restricted to the 93 news sites in the sample, against roughly 5% for every other sector combined.
Key findings
The sharpest split runs along language lines within the news sector. Some 86% of French-language news sites block at least one AI crawler, and 82% specifically block the most widely used AI training crawler. The comparable figure for English-language outlets is 39%. One Quebec media group has drawn a distinction between training and retrieval: it blocks AI training crawlers while leaving AI search access open, an approach that is technically possible via robots.txt directives but requires deliberate configuration.
Broader inconsistencies complicate the picture. Nine sites, including several major broadcast newsrooms, still list crawler names that the AI companies behind them no longer use, leaving current crawlers without any restriction. A separate finding shows 16 sites block the crawler of a Canadian AI developer that received a $240 million federal funding commitment in 2024, among them some of the country's largest national newsrooms. One national newspaper chain takes a different approach entirely, routing 19 publications to a machine-readable content licence rather than a flat block.
Jeremy McWilliams, founder of Chateau Mojo, said: "When most of Canada's journalism asks AI to stay out while banks, retailers and governments let it in, what AI tools can freshly read about the country tilts toward the people selling something."
Market and regulatory context
The study lands in the middle of an unresolved global argument about whether training and retrieval crawling requires a licence, revenue share or other commercial arrangement with publishers. Several European and US news organisations have pursued litigation or licensing negotiations with major AI developers, and Canada's Online News Act has already restructured how platforms pay for news links. The robots.txt mechanism is voluntary and relies on crawler compliance; it carries no legal force under Canadian law, and the study explicitly notes it does not capture network-level blocking.
For financial services and government, the data suggests a different calculus. Banks and insurers appear to have concluded that AI visibility is commercially useful: an AI assistant unable to read a firm's site cannot recommend its products. That logic may shift as regulators pay closer attention to how AI tools surface financial advice and government information. The EU AI Act's transparency provisions and the UK's ongoing AI and copyright consultation both point toward tighter oversight of how training data is sourced and disclosed.
Chateau Mojo has made the full dataset, including the name of every site and every crawler rule examined, freely available with attribution. The study covers what sites ask crawlers to do rather than what crawlers actually do, and should be read accordingly. A follow-up cut comparing year-on-year change in crawler policy would materially strengthen the findings; the consultancy has not indicated whether one is planned.