Core processing stays in the browser
When you use the current inspection and cleanup tools for JPEG, PNG and WebP, parsing and cleaned-file generation happen in the browser. The source image does not need to be uploaded to a MediaClean server for those functions.
The browser creates temporary local object URLs for previews and output downloads. Those URLs belong to the local browser session and do not represent public image hosting.
What can the page request separately?
Like any website, the page itself may make technical requests for code, fonts, static files or disclosed performance analytics. That is separate from the image content processed locally by the current cleaner.
If a future feature requires a server or external processing provider, we will not describe it as browser-local. The interface will disclose that distinction before the feature is used.
A practical privacy recommendation
Keep the original when you need metadata for rights, archives or workflow recovery. Use a cleaned sharing copy instead of overwriting your only source file.
For highly sensitive images, review the post-cleaning scan yourself before publishing or sending the output.