In the academic area, recommender systems have received a lot of attention in the recent years (see for instance the increasing success of the RecSys conference). This success has also reached industry and the general public via large platforms such as Amazon or Netflix. The recommender systems present multiple benefits that can be attributed to automation by means of the migration of function from the operator to the Recommender System itself. Depending on the type of recommender system, these functions can cover: support to perception of information,support to identification of potentially relevant elements, support to the selection of one element amongst a list... Despite their widespread use, their development has not reached the level of maturity required for the deployment in the context of critical command and control systems. This position paper identifies some pre-requisite for making recommender systems deployable in critical context of critical systems.