The page shows a crop.
The file holds the whole thing

How cropping works in a layout

When you place a photograph into a document and drag its handles to crop, most layout software does not cut the image. It stores the original and records a clipping region saying which part to display. This is sensible: it lets you adjust the crop later without losing anything. It also means the export carries the complete photograph, with the page displaying a window onto it.

Which is a different problem from the thumbnail

This is a separate mechanism from the embedded preview described in cropping is not deleting. There, an old thumbnail survives inside the image file. Here, the full-resolution image is deliberately retained by the document and simply not shown. Both have the same consequence and neither is visible when you read the document.

What it means in practice

A property listing that crops a neighbour out of a photograph. A report where a screenshot was cropped to hide a column of figures. A slide where a face was cropped from the edge of a group photo. In each case the original is recoverable from the exported PDF by anyone who extracts the images, which takes seconds and no expertise.

Extraction is lossless, which cuts both ways

Photographs are stored inside a PDF as complete JPEGs. Pulling one out means copying that stream to a file: no decoding, no re-encoding, nothing lost. That is useful when you want an image back at full quality from a document someone sent you. It is equally useful to anyone curious about what a page is not showing them.

Crop before you place

The fix is straightforward and takes a moment. Crop the image in an image editor, export it as a new file, and place that file into the document. Then the document has only the part you meant to include, because the rest no longer exists in anything you handed over.