understanding request for filtered content. the "sensitive content" marker is too general.

suggestion

- make multiple type of sensitive content flag

- dont flag the words. just the image