Most websites' AI bot instructions go completely unread

Ahrefs' analysis of 137,000 domains reveals that the llms.txt protocol—meant to guide how AI systems crawl and use website content—is almost entirely ignored in practice, with 97% of files receiving zero requests from AI bots. Websites are creating these files to appear responsible, while AI companies' crawlers largely bypass them, leaving the protocol functionally useless as a control mechanism. For publishers and brands worried about content scraping, protection will come through legal leverage, technical barriers, or direct deals with major AI labs—not voluntary machine-readable instructions.