Tanguy Ortolo

With the PRISM scandal, there has been some talk about encrypted communication systems. For instance, BitMessage is often introduced as an easy and secure message system, that would allow you to communicate with no possible eavesdropping. Apple is also making similar claims about their systems iMessage and FaceTime.

This is a good time to remind this: without direct contact or a certification layer, encryption systems are not secure! Or at least, not as secure as you would expect, as they do allow some kind of eavesdropping.

Let us take the example of BitMessage:

1. Alice sends her BitMessage address to Bob by email, an insecure channel;
2. Mallory catches that email message and changes Alice's BitMessage address with his own;
3. Bob sends a BitMessage for Alice to the address he received, which he thinks is Alice's, but is in fact Mallory's;
4. Mallory receives the message, reads it, modifies it if he wants, and sends it to Alice.

This is a man-in-the-middle attack, a kind of active eavesdropping technique that requires read-modify access to a communication line. Without a specific way to ensure that a cryptographic key really belongs to its alleged owner, a cryptographic system is vulnerable to such attacks, even if it does protect against simpler attacks (like passive eavesdropping or introduction of MiTM after the introduction sequence). When you read that a cryptographic system provides end-to-end security and is impervious to eavesdropping, while it provides no mean to make sure you are in contact to the right person, remember to consider this affirmation as what it is: a lie.