I see some criticisms (that I agree with) of the "security" model here.

Implementing the self destruction protection on the client is problematic (even in the case of SnapChat) because the end user is in control of the client. So if users are motivated enough they'll be able to circumvent any protections you implement.

The nice thing is that most users are not motivated enough to do what's necessary to retrieve the self-destructing messages. It's a lot easier to save the message when sent to a web client but it's still beyond the ability of most users.

So no these applications aren't providing real security but it's close enough for the types of messages that I assume people are sending.