Product transparency

Where your data goes

Image files and extracted metadata are processed on your device during normal tool use. They are not uploaded by the tools. Ordinary page and asset requests still reach our hosting/CDN infrastructure.

Selected image → Browser / workers → Local result → Optional download

Website requests → Hostinger / hCDN → Page and code

Selected files go to local processing code and results stay on the device. Separately, the browser requests website pages and assets from the hosting/CDN infrastructure.

On your device

Image bytes, decoded pixels, filenames, extracted metadata, GPS, AI prompts, embedded provenance and generated outputs are handled in browser memory. Tools can hand files to another tool using an in-memory reference. The URL token contains no filename or extracted metadata, but the URL itself can reach hosting and browser history.

Website infrastructure

Hostinger and hCDN deliver the site and its assets. Ordinary requests disclose IP addresses, paths, timestamps and browser/request headers. Processing locations: wherever Hostinger and its CDN operate infrastructure, which can be outside your country. Hostinger keeps server and security logs under its own retention practices. MetaPurify runs no analytics or log database of its own.

Hostinger and hCDN may use their own service providers, as described in Hostinger’s privacy policy.

Storage, cookies and origins

The application sets no cookies and uses no localStorage, sessionStorage, IndexedDB, Cache Storage or service worker. Browser HTTP caching, history and user-requested downloads are separate browser functions. Object URLs provide temporary local previews and downloads.

Runtime scripts, styles, fonts, workers and codecs are served from this website’s origin. There are no third-party runtime scripts, embeds, behavioral analytics or ads. External reference links are navigations you choose to follow. The C2PA checker disables remote-manifest and online certificate-status fetching.

Claims and regression checks

Image content is processed locally: per-tool browser journeys inspect request URLs, headers and bodies for sentinel filenames and metadata and reject unexpected origins or write requests.
No persistent application image storage: journeys check cookies, local/session storage, IndexedDB and Cache Storage, and instrument storage writes.
Temporary object URLs are released: tests check cleanup after leaving each tool.
Public statements stay consistent: release validation and legal-route tests check required facts and reject internal publishing notes.

These describe the regression suite, not certification or proof for every possible file. The release audit records results separately for the local production build and the deployed site. A homepage check alone does not verify every tool.

Limitations

Supported metadata structures and formats vary by tool. C2PA checks do not establish scene authenticity or a trusted signer identity, and remote-only credentials are not retrieved. Visible image content, invisible watermarks, proprietary structures, browser extensions, compromised devices and later sharing remain outside the local-processing promise. Resource limits reduce but cannot eliminate memory exhaustion risks on all devices.

Automated accessibility checks do not establish complete WCAG conformance. See accessibility, privacy and terms.

Change history

2026-09-22: removed a statement about hosting-log access that had not been verified with the host.

2026-09-21: added separate file/infrastructure flows, explicit C2PA network settings and per-tool privacy regression coverage.