A *secret key* in cryptography is supposed to be just that -- something that is kept completely secret.  However, there is a growing realization that in reality a secret key is not totally secret, as a variety of so-called "side channel" attacks can uncover some information about the secret key, such as some of its bits.  
 
In this talk, I will survey some recent results on cryptographic primitives that are provably secure even with keys which may be partly compromised. I will focus on a recent joint work with Zvika Brakerski, Jonathan Katz, and Vinod Vaikuntanathan, which shows how to construct cryptographic schemes that are secure against ``continual leakage''.