Abstract: Recently, Danielson, Satishchandran, and Wald proposed a novel mechanism for decoherence in spacetimes with Killing horizons. When Alice performs a Stern-Gerlach type experiment outside the horizon, there is a fundamental rate of decoherence due to the emission of soft gravitons/photons that radiate away information. In this talk, I will sharpen and generalize this mechanism by recasting the gedankenexperiment in the language of (approximate) quantum error correction. By explicitly computing the quantum fidelity between the vacuum state on the horizon and the state with soft radiation propagating through the horizon, we characterize how much an eavesdropper (Eve) behind the horizon is able to learn about Alice’s experiment. From the information-disturbance tradeoff, there is a direct relation between the information gained by Eve and the decoherence of Alice’s generalized experiment. I will comment on how this phenomenon can be interpreted as a low-energy probe of the so-called central dogmas of black hole and cosmological horizons.