Most websites actively shut the door on large language models. That is the central finding of an AI visibility study and audit published by Stradiji, a result with direct consequences for how brands appear inside AI-generated answers.

The audit examined how websites present themselves to LLM crawlers and concluded that the majority effectively block them. In practice, that means chatbots and AI search tools cannot read, quote, or recommend those sites when users ask questions.

What does the study actually measure?

Stradiji's audit looks at AI visibility: whether a site's content can be picked up and used by language models. A site can rank well in conventional search and still be absent from AI answers, because the two systems read the web differently.

The mechanism is simple. LLMs rely on crawlers fetching pages. When a site's robots rules, technical setup, or content structure block or confuse those crawlers, the model never sees the content. The site exists for human visitors but not for AI tools.

What does this change for site owners?

For operators, the finding turns a quiet technical setting into a commercial decision. If most sites block LLMs, the few that allow access gain a comparatively larger share of AI citations and recommendations. Visibility inside AI answers is becoming a distribution channel in its own right, alongside traditional search rankings.

The audit suggests the problem is widespread rather than isolated. Fixing it typically involves reviewing crawler access, page structure, and how content is exposed to non-human readers.

Why this matters now

AI tools increasingly act as the first stop for product research and questions that once went to a search box. A site that is invisible to those tools loses that traffic before a competitor is ever compared on price or features.

Stradiji's study frames the issue as a gap between awareness and action. Site owners who assume AI models will find them by default may discover the opposite: by configuration or neglect, most sites have closed the door.

The practical takeaway is an audit of one's own setup. The study indicates the majority have work to do.

By Sophie Lindqvist