An example is https://ca.wikipedia.org/w/index.php?title=Smith_%26_Wesson&curid=940802&diff=30407429&oldid=30407426
This was anticipated in https://gerrit.wikimedia.org/r/c/mediawiki/services/parsoid/+/804404
VE will display the content in both the alt field and the caption, and they'd need to be edited in sync to avoid being duplicated and diverge. The editing experience when the caption isn't visible can stand to be improved as it is though.
Opening this task to track how big of an issue it becomes ...
The UI in VE should probably have a checkbox that says "Use the caption as the alt" and have it checked by default. It'll be unchecked if the caption and alt differ in the source. And note that all this only applies if the media isn't a thumb/manualthumb/frame.