The notion of ``boundary ambient'' has been recently introduced to model multilevel security policies in the scenario of mobile systems, within pure Mobile Ambients calculus. Information flow is defined in terms of the possibility for a confidential ambient/data to move outside a security boundary, and boundary crossings can be captured through a suitable Control Flow Analysis. We show that this approach can be further enhanced to infer which ambients should be ``protected'' to guarantee the lack of information leakage for a given process.