How Recall uses AI

What the AI does

When you save something to Recall, AI does the work that makes it useful later. For every save it can:

• write a short summary and pull out the key points;
• tag it and classify it by platform and content type;
• identify the people, places and topics inside it;
• make it searchable by meaning, so you can find it by describing it;
• answer questions about it — in the Ask tab across your whole library, and in Chat with Bookmark for a single save.

Without AI processing, Recall would just be a list of links. The AI is what turns it into a library you can actually search and question.

What happens to your saves

To do this work, the content you save — the URL, text, transcripts and any images — is sent to Anthropic, the company behind the Claude AI models, through its secure API. Claude reads the content, and Recall stores the results (the summary, tags, and so on) in your library.

Anthropic is the only AI provider Recall uses. We do not use OpenAI or other AI services. A few non-AI providers help extract content from links before it reaches Claude — these are listed on our sub-processors page.

Retention and training

• Anthropic keeps the content in its API logs for a limited period — up to around 30 days — for safety and abuse monitoring, then deletes it.
• Your saves are not used to train Claude or any other AI model.
• The results that stay in your library (summaries, tags, transcripts) remain until you delete the save or your account.

Your control

Because Recall is built around AI, AI processing is part of the core service and cannot be switched off on its own. You stay in control in two clear ways:

• You choose what to save. Nothing is processed unless you save it. If you don't want something analysed by AI, don't add it to Recall.
• You can stop entirely. Deleting your account, in Profile → Privacy & AI, ends all AI processing and removes your library. See the account deletion page.

There is also a separate, optional setting — "Share anonymised usage" — that lets you consent to anonymised prompts and responses being used to improve Recall's own AI quality. It is off unless you turn it on, has no effect on the core AI processing above, and can be switched off again at any time.

A note on accuracy

Where to read more

For the full picture of how your data is handled, see our Privacy Policy and sub-processors page. You can also read Anthropic's privacy policy for how Anthropic handles data sent to its API.