We provide a set of semantically secure achievable rates for the fast fading wiretap channel. In particular, we do so for the cases where there is channel state information at the transmitter (CSIT) for both the main and eavesdropper channels (full CSIT), for only the main channel (partial CSIT), and for neither channel (statistical CSIT). In the case of partial CSIT and statistical CSIT fast-fading channels, we show that this coding scheme can achieve the best known achievable rates. In the case of full CSIT fast-fading wiretap channels, we show that this coding scheme can actually achieve the secrecy capacity. In particular, this implies that the semantic secrecy capacity for these channels is equivalent to the weak and strong secrecy capacities. Moreover, we achieve these rates in a way that is non-invasive to existing systems and also happens to be explicitly given as well as efficient in implementation.