We sketch a relation between two crucial, yet independent, fields in quantum information research, viz. quantum decoherence and quantum cryptography. We investigate here how the standard cryptographic assumption of shielded laboratory, stating that data generated by a secure quantum device remain private unless explicitly published, is disturbed by the einselection mechanism of quantum Darwinism explaining the measurement process by interaction with the external environment. We illustrate the idea with a paradigmatic example of a quantum random number generator compromised by an analog of the Van Eck phreaking. In particular, we derive a trade-off relation between eavesdropper's guessing probability $P_{guess}$ and the collective decoherence factor $Γ$ of the simple form $P_{guess} + Γ\geq 1$.