The security policy of an information system may include a wide range of different requirements. The literature has primarily focused on access and information flow control requirements and more recently on authentication and usage control requirements. Specifying administration and delegation policies is also an important issue, especially in the context of pervasive distributed systems. In this paper, we are investigating the new issue of modelling intrusion detection and reaction policies and study the appropriateness of using deontic logic for this purpose. We analyze how intrusion detection requirements may be specified to face known intrusions but also new intrusions. In the case of new intrusions, we suggest using the bring it about modality and specifying requirements as prohibitions to bring it about that some security objectives are violated. When some intrusions occur, the security policy to be complete should specify what happens in this case. This is what we call a reactio...