When asked how they might deal with abuse of the service to distribute illegal files, he suggested that you could compare uploaded files to hashes of known files. This doesn’t make sense in a system where the server has no knowledge of the unencrypted file, since the same file encrypted with two different passwords will result in two different hashes.
The only way I could see them flagging potentially illegal files on the server-side if they don’t have access to the cleartext file would be through the filesize, and that would lead to too many false-positives. On the client-side it could be done through a local checksum against a denylist (compared locally for privacy reason) before uploading, but that could be easily defeated.
One red flag from that podcast:
When asked how they might deal with abuse of the service to distribute illegal files, he suggested that you could compare uploaded files to hashes of known files. This doesn’t make sense in a system where the server has no knowledge of the unencrypted file, since the same file encrypted with two different passwords will result in two different hashes.
Can’t you hash it before uploading and upload just the hash? Or download the banned hash list locally.
Sure, but then you’re trusting the client. I can always encrypt
x
and send along the hash fory
.In the end you can always just encrypt the illegal stuff externally before giving it to them…
The only way I could see them flagging potentially illegal files on the server-side if they don’t have access to the cleartext file would be through the filesize, and that would lead to too many false-positives. On the client-side it could be done through a local checksum against a denylist (compared locally for privacy reason) before uploading, but that could be easily defeated.