<p>Injection-induced seismicity represents a major challenge for the development of Enhanced Geothermal Systems (EGS). To effectively mitigate the associated seismic hazard, a better understanding of the spatiotemporal evolution of induced seismicity and its efficient modeling are required. Towards that end, a stochastic framework within the continuous time random walk (CTRW) theory is used to make inferences regarding the diffusion properties of injection-induced seismicity in three cases of hydraulic stimulations in EGS. The analysis of seismicity within the CTRW context indicates multi-scaling variations in the waiting times distributions and in the evolution of the mean squared distance of seismicity with time, both associated with the co- and post-injection periods, respectively. During fluid-injections, an almost Poissonian waiting times distribution is followed by broad distributions during post-injection, enhancing long-term clustering effects and inter-earthquake interactions. At the same time, the rate of triggered earthquake diffusion drastically drops during the post-injection period for all the studied cases. Such properties may have implications on the main driving mechanisms of injection-induced seismicity in EGS, highlighting the transition from a dominant pressure-driven triggering mechanism during fluid-injections, to a mixed mechanism after termination of injections, where stress transfer effects and inter-earthquake interactions become more important.</p>