AI Content Signals

AI Content Signals lets you declare, in a machine-readable way, how AI systems may use your content: for search indexing, real-time AI answers (RAG), or model training. It started with Cloudflare’s Content Signals in robots.txt and now expresses the same preferences across several surfaces, so more crawlers and tools can read them:

• robots.txt — Cloudflare Content Signals (search / ai-input / ai-train)
• HTTP header — X-Robots-Tag: noai, noimageai
• HTML meta — robots noai, noimageai
• /.well-known/tdmrep.json — W3C Text and Data Mining Reservation Protocol
• EU Directive 2019/790 rights reservation

Everything is opt-in and does not change your robots.txt output unless you enable it.

The signals you control

You set three preferences, and AI Content Signals expresses each one in the right format on every surface you enable:

• search — allow or deny search indexing and traditional search results (links and short snippets)
• ai-input — allow or deny using your content for real-time AI answers (RAG, grounding, AI Overviews)
• ai-train — allow or deny using your content to train or fine-tune AI models

These three come from Cloudflare’s Content Signals vocabulary, written to your robots.txt. When you deny AI training, the same opt-out is also emitted on any other surface you enable: noai, noimageai in an HTML meta tag and an X-Robots-Tag header, plus a tdm-reservation in your TDMRep manifest. Declaring the same preference in several places means a crawler that ignores one signal may still honor another.

Key Features

• Easy-to-use settings page in WordPress admin
• Set global defaults for all crawlers
• Configure specific settings for individual AI bots (GPTBot, ClaudeBot, PerplexityBot, etc.)
• Add custom bot User-Agents
• Supports both physical and virtual robots.txt files
• Option to create physical robots.txt with basic WordPress rules