We examine secure communication in multi-receiver models when the eavesdropping terminal(s) are capable of tapping into a subset of the transmitted codeword(s) of their choice. This generalizes classical wiretap models into those with a more capable adversary. The point-to-point channel with such an adversary was presented last year along with its strong secrecy capacity. Among the models we examine this year is the broadcast wiretap channel (B-WTC) where the adversary eavesdrops on the legitimate communication and is additionally given access to a subset of the symbols of her choice. Additionally, we introduce new models for the broadcast and interference channels with confidential messages, (BC-CM) and (IC-CM), where each receiver, besides his noisy observations, is provided with a subset of noiseless observations for the transmitted codeword(s). We present achievable strong secrecy rate regions for these three models. Achievability is established by solving for dual multi-terminal source models, and requires extending the existing tools to multiple correlated sources. For each model, we highlight the insights drawn from our analysis, along with optimality results for special cases. In general, for the new B-WTC, the derived rate region indicates the secrecy cost due to the additional capability at the wire-tapper; for the new BC-CM and IC-CM models, the size of the subset at each receiver induces a trade-off between their rates.