What Data Does the AI PDF Summarizer Send to the Provider?

See the exact privacy boundary: local PDF extraction first, explicit consent, up to 30,000 document characters, recent chat messages, and direct browser-to-provider requests.

Review AI PDF Summarizer

The source PDF file is processed locally first

The current component passes the selected PDF into 1into1’s local text-extraction pipeline. Digital page text is read in the browser, and low-text or scanned pages can use the local OCR path.

The resulting text is kept in application memory for the AI workflow.

Nothing is sent until the consent box is checked

If consent is not checked, the send function stops and shows a notice instead of contacting the selected AI provider for the chat workflow.

Consent is tied to the current file and destination choice: selecting another PDF, provider or custom endpoint clears it.

The provider receives a text excerpt and recent conversation context

The system prompt contains up to the first 30,000 characters of extracted document text. The chat payload also includes up to the six most recent messages from the current conversation.

The PDF binary itself is not attached to that chat request by the current component.

The request goes directly from the browser to the configured endpoint

The component uses browser fetch calls against the provider URL or the custom endpoint entered by the user. Authentication is sent as a bearer token when an API key is present.

Because the destination is a third party or your own server, its own privacy, retention, security, rate-limit and billing rules govern what happens after the request reaches it.

API key storage is separate from provider preferences

The third-party API key is written to session storage, while the selected provider, custom model and custom endpoint can be stored in persistent browser local storage.

MDN notes that session storage belongs to the current page session, while local storage persists across browser sessions unless it is cleared or browser policy removes it.

Official sources reviewed

  1. MDN — Using the Fetch APIExplains request URLs, methods, headers and bodies for browser fetch requests.
  2. MDN — Web Storage APIExplains the difference between sessionStorage and localStorage used for key and preference storage in the browser.