- You upload an image of a cute cat to your webserver
- You post the link to the image on some social network (using a throwaway account)
- You send the link to the post to a friend, who likes the post
- You switch out the image on the server for a photo of hitler (url stays the same)
- You wait a day for all the caches involved to update
- You act disgusted and ask your friend wtf they were thinking when they liked that image
The remarkable thing is that it still works in the walled garden of social networks like Facebook, even though their main value-add is that they can stop people abusing the system like this.
- Everyone shrugs, says "computers eh?" and moves on with their lives
This attack would only work for gaslighting or repeated targeting I think, and even then, I don't think it would be that convincing.
You forgot the step of
-take an image of the transgression and post it to Twitter
-50000 other people see you liked hitler.
-your next job sees this image on the internet, so actually no, you dont get that job.
But is it the author editing the post who change it or what?