Content Signals
Content Signals are lines in robots.txt that say how your content may be used once it's read: for search, as AI input, or for AI training.
Content Signals are lines you add to your robots.txt to say how your content may be used after a bot has read it. robots.txt has always said which pages a bot may read. Content Signals add what it may do with them: build a search index, feed an AI answer, or train an AI model.
Cloudflare published the Content Signals Policy in September 2025, under a CC0 licence, so anyone can use it.
The three signals
There are three, and you set each to yes or no:
search: building a search index and showing results, such as links and short excerpts. This doesn't include AI-written summaries.ai-input: feeding your content into an AI model to answer a question, such as for AI search answers.ai-train: training or fine-tuning AI models.
If you leave a signal out, you haven't said either way.
What it looks like
For example, Fernhill wants to appear in search results and in AI answers, but doesn't want its writing used to train models. Its robots.txt says:
User-Agent: *
Content-Signal: search=yes, ai-input=yes, ai-train=no
Allow: /Every bot may still read every page. The signal adds what Fernhill allows them to do with what they read.
What it can and can't do
Like the rest of robots.txt, Content Signals state a preference. They only work if the companies reading them choose to follow them. They're most useful when one company uses the same bot for several jobs, because a plain block can't keep the bot out of training without keeping it out of search too.
In AnswerOrbit
The Audit tab in Website reads your robots.txt every day, and History keeps every version we've read, with the lines added and removed each time. See Website and AI traffic.