A model for the spreading of online information on multiplex networks is introduced and analyzed using branching-process methods. We demonstrate that the dynamical system is, in the limit of low innovation, poised near a critical point, which is known to lead to heavy-tailed distributions of popularity similar to empirical data.