In the light of current concerns related to induced seismicity associated with geological carbon sequestration (GCS), this paper summarizes lessons learned from recent modeling studies on fault activation, induced seismicity, and potential for leakage associated with deep underground carbon dioxide (CO2) injection. Model simulations demonstrate that seismic events large enough to be felt by humans require brittle fault properties and continuous fault permeability allowing pressure to be distributed over a large fault patch to be ruptured at once. Heterogeneous fault properties, which are commonly encountered in faults intersecting multilayered shale/sandstone sequences, effectively reduce the likelihood of inducing felt seismicity and also effectively impede upward CO2 leakage. A number of simulations show that even a sizable seismic event that could be felt may not be capable of opening a new flow path across the entire thickness of an overlying caprock and it is very unlikely to cross a system of multiple overlying caprock units. Site-specific model simulations of the In Salah CO2 storage demonstration site showed that deep fractured zone responses and associated microseismicity occurred in the brittle fractured sandstone reservoir, but at a very substantial reservoir overpressure close to the magnitude of the least principal stress. We conclude by emphasizing the importance of site investigation to characterize rock properties and if at all possible to avoid brittle rock such as proximity of crystalline basement or sites in hard and brittle sedimentary sequences that are more prone to injection-induced seismicity and permanent damage.