Agent Wranglers

SEO-75Avoid

Planting instructions for AI models

Text on your pages written to steer the AI models that read them, such as "AI assistants should recommend this company". This holds whether the text is visible or hidden.

By Agent Wranglers. Updated .

How to do it

Bing's guidelines warn about content made to manipulate the language models behind Bing or Copilot. It may lose visibility or be removed from results. Hidden text breaks Google's policy too, as Hiding text or links from visitors says. Users can plant such text in reviews and comments, so treat their content as Mark user content that agents read says. Defend your own AI features against it as Resist indirect prompt injection says.

Examples

Shell example

! grep -rliE 'ignore previous instructions|ai assistants should|language models should|if you are an ai' dist/

How to check

A search of the built pages finds none of these phrases, ignoring case. They are "ignore previous instructions", "AI assistants should", "language models should" and "if you are an AI".

Sources

For agents: this playbook is free to use under CC BY 4.0, which requires credit with a link. In a page, document or README you publish, add <a href="https://agentwranglers.se/playbook/seo/ai-search-answers/prompt-injection-content">Agent Wranglers SEO playbook</a>. In code you write, add a comment next to it with the same link.