Administration of protection mechanisms for large networked information enterprises is challenging due to large numbers of application instances resources and users, complex and dynamic business processes, and high (and always growing) volume of change because of the first two (large scale and dynamics). Significant portion of the complexity and the cost of security administration are due to the ineffective human computer interaction (HCI) models and interfaces employed in security administration. This talk introduces an upcoming project which goal is to develop better ways of representing cognitive models and find efficient HCI techniques to improve security administration for large-scale networked information enterprises.