Security

Document Parser is designed around local processing. When you choose a file, the app reads it in the browser, attempts to extract selectable text directly when that text already exists, and only falls back to OCR when a page or image does not contain usable text. In practical terms, that means PDFs with embedded text can often be parsed quickly, while scanned pages and image-based files may require OCR work inside the browser before readable output is produced.

The extracted text remains in the current browser session until you copy it, download it, or start over, and the app does not require a backend round-trip to perform its core parsing flow. This security model assumes the current client-side deployment of the app, so if anyone adds analytics, external APIs, logging, cloud storage, or other integrations later, the security and privacy expectations would change and this page should be updated to reflect that new behavior clearly.

Use trusted browsers, keep your device updated, and avoid processing sensitive files on shared, public, or unmanaged machines where browser data may be exposed to other users. If you plan to move extracted text into an LLM, embedding workflow, or retrieval system, remember that the privacy and security posture of the overall workflow will then depend not only on this app’s local parsing model, but also on whatever downstream tools, vendors, storage layers, and access controls you use after the text leaves the browser.