There was a great discussion I wont be able to find about this recently on HN after an article about a zebra/leopard patterned couch. Thwarting an image id algo

There was a great discussion I wont be able to find about this recently on HN after an article about a zebra/leopard patterned couch. Thwarting an image id algorithim is not difficult and the researcher came on and spoke about the problems in the space similar to this xkcd[0] comic. It is a super complex problem to solve. How far away are we? It is like the turing test, everytime someone gets close the term is redefined. It is certainly coming, maybe reasonable models in ~5 years publicly acknowledged. What can someone really do to prevent it? Short of making an account every few months with a lot of your data but different pictures, what american doesnt have a photo that has been published on the internet? Even if you dont have facebook, someone you know does. The tech will take a while to dev and refine, and longer to deploy at scale, but I can't think of a scalable solution to this. [0]http://xkcd.com/1425/ http://xkcd.com/1425/