As more robots, such as autonomous vehicles, are deployed in life-critical situations it is imperative to consider safety, and in particular, localization safety. While it would be ideal to guarantee localization safety without having to modify the environment, this is not always possible and one may have to add landmarks or active beacons for landmark-based localization. As such, this work introduces a method to identify the minimum places in an environment where landmarks can be added to ensure localization safety, as quantified using integrity risk, the probability of undetected sensor errors causing localization failure while accounting for measurement faults. The letter formulates the problem as a systematic minimization: given the robot's trajectory and the current landmark map, add the minimum number of new landmarks such that the integrity risk along the trajectory is below a given safety threshold. The letter proposes three algorithms: a naive approach, Integrity-based Landmark Generator (I-LaG), and Fast I-LaG. The computationally expensive naive algorithm serves as a reference to illustrate simple scenarios. I-LaG adds relatively fewer landmarks than the Fast I-Lag algorithm but is more computationally expensive. Simulation and experimental results validate the proposed algorithms.