Shadow Content for AI Search Repeats Old SEO Mistakes

Old SEO Lessons Meet New AI Shortcuts

This essay questions a growing habit in AI search work: creating content meant for agents rather than people, kept apart from the public website. The author traces how search engines learned to punish manipulation such as hidden text, doorway pages and cloaking, and argues that the same pattern is now forming around llms.txt files, agent-specific markdown and dedicated endpoints. The useful distinction it draws is between presenting the same information in a machine-friendly format and publishing different information for different audiences.

For brand builders, the piece is a sharp check on shortcuts that look efficient on paper. It is candid that many of these approaches are legitimate, and that structured data and feeds have always been valuable. Anyone responsible for content governance, technical SEO or AI visibility should read it, because the real risk it names is a second, less visible content estate that is hard to keep accurate and trustworthy alongside the main site.