Currently when shared on social media and elsewhere, the main OCR tool website at ocr.wmcloud.org does not expose its image or description:

(This example is as rendered on Mastodon, but it's similar elsewhere.)

To fix this, it should have Open Graph metadata added, with the existing logo used for image.

The description is currently

Transcribe text from images

but should perhaps be modified to

A tool to transcribe text from scanned images on Wikimedia Commons, for use on Wikisource and elsewhere.