> Just to highlight one of the issues, if you're using a webapp, the server itself is what tells your browser what code to run. So, if the webapp can display de

> Just to highlight one of the issues, if you're using a webapp, the server itself is what tells your browser what code to run. So, if the webapp can display decrypted message content in a browser, a hacked server can just deliver JS code that causes the browser to fetch every single message of history your browser client has access to and send them back to the server unencrypted. If a native app is compromised, you are screwed all the same. I don't see how the delivery channel makes a difference -- the server serving the web client doesn't necessarily have to be the same one that's relaying the messages...