The visual state of an image no longer changes when clicking on it. There's no visual clues such as borders, drag handles or anything that denotes that the image has been left clicked. Therefore it is very easy from a UX point of view to delete the image by atempting to type some text.
Visual clues to show that the image is currently selected, a border or an opacity fade added to the selected image
No visual clues are given as to the selected status of the image