How is Cameron going to ensure that law enforcement can read all communications? One way would be to provide systems with ”back doors”; introducing intentional vulnerabilities. We all know that won’t work. Or rather will work much better than intended, if you get my drift.
Some, including Steve Gibson, maintain that it can in fact be done by having law enforcement maintain a secret, well-guarded, key and mandating that all messages sent are including that encryption target in every message. That would allow LE to decrypt it using a very carefully guarded secret key, if need be. All this without weakening the actual encryption mechanism.
The problem with this is that LE can’t know if everyone is following the law without actually trying to decrypt messages flying by. And to do that on a large scale by necessity implies that the “highly guarded” secret key must be available on a large number of systems, exposing it to compromise.
Even if we stipulate that there is some, hitherto unknown, mechanism that allows LE to verify that messages in fact include the LE destination without having the secret key available, they still can’t know if the encryption is valid until attempted. For instance, the encrypted symmetric key may be intentionally wrong. Or, the encrypted message may contain another encrypted message which does not contain the LE mandated item. And that, in turn, can only be discovered once you perform the actual decryption, which requires the ”highly protected” government key.
In other words, it won’t work.
2 thoughts on “How even key escrow won’t work for Cameron”
I’ve wondered–what Steve was proposing required two payloads in the data.
One that was encrypted with the ‘proper’ key, and another equally large payload encrypted with the ‘secret government’ key.
Wouldn’t that place an enourmous strain on the network, and that you’d double the network load overnight due to government policy as opposed to engineering efficiency?
The payload is always encrypted with a random symmetrical key, for efficiency. Then that key is in turn encrypted with one or several public keys. What he proposed was to add the government’s public key to that. Exactly the same mechanism as we use when encrypting to several receivers normally. Doing it this way only increases the total with up to a few hundred bytes, regardless of the size of the actual payload, so that’s not the problem.