The general media has (temporarily) discovered hashing. I predict a short run of calls for social media platforms to use it to prevent the spread of hate videos, violent videos, revenge pr0n, etc, etc, etc.
I've seen hashing in use for some time. Fifteen years ago it was very popular as the increase in the number of viruses exploded. Not so long ago Facebook tried using it in an odd, rather futile, and foolish attempt to prevent revenge pr0n. It's been used to prevent the theft of music and video as intellectual property for some time.
It works, a bit, but not terribly well.
The idea is to detect something you don't want spread, and then take a hash of it. You can then search, relatively quickly, and compare that hash value against the hash values of either existing files, or newly uploaded files (depending upon your application).
I said "relatively" quickly. One of the people quoted in that article said "It's exceedingly fast." It's exceedingly fast compared to more detailed forms of analysis. But when around 10 hours of video are uploaded to YouTube alone every second (anybody have current statistics?) ... well, hashing does take some time, and little bits add up. And then there is the time to compare every hash against every other hash ...
And hashing only works if nothing has been changed. After all, hash values are used, sometimes in digital signatures or certificates, to ensure that something hasn't changed. Again, someone in the article referred to "'robust' hashing — a method that should be able to detect variations on reuploads." That's an interesting use of the word "robust." I'd think most people in the crypto field would think of a "robust" hash as one that would detect any changes, not one that would allow some changes and still match. But, quite aside from the use of the word "robust," making a hash that will accept some changes and still detect "similar" is a non-trivial task. And such a hash function would likely take even more time to run.
It's easy to use hashes to catch direct and identical copies. But videos can be modified in all kinds of ways. They can be edited for length, cut into collections, processed to add comments, or even just drop a few packets during streaming. Any or all of these events could mean that a hash value will not match.
No, I don't think hashing will be the silver bullet people are looking for ...
............
Other posts: https://community.isc2.org/t5/forums/recentpostspage/user-id/1324864413
This message may or may not be governed by the terms of
http://www.noticebored.com/html/cisspforumfaq.html#Friday or
https://blogs.securiteam.com/index.php/archives/1468
