Among the different technological solutions realized in order to preserve data privacy, the front end trust filter could be effectively applied in environments characterized by high dynamism and untrustworthiness. Unfortunately, a preliminary assessment of this approach suggested a possible weakness: by using different user’s profiles the privacy policy can be eluded and sensitive information could be obtained by inference over legal data set. This paper proposes a solution that could be helpful for two purposes: it could be used in the design phase for identifying which use scenarios (i.e., sequences of legal queries) could threaten data privacy; additionally, it could be used for identifying users which could potentially exploit inference for disclosing confidential information.