Access control systems depend on the administrator formulating sets of policies determining the conditions for access. These policies can include many rules, and can change over time: additions, modifications and deletions can be performed. When administrators perform these operations, they can unintendedly compromise the integrity of the policy sets. Other problems that can occur are related to possible incompleteness or to rule overlap. Incompleteness, inconsistency and other potentially harmful policy integrity problems will be called ‘semantic anomalies’. We will demonstrate a prototype Policy Assistant that flags anomalies in sets of EEM policy rules, thereby assisting administrators in maintaining their integrity . The same principles used for the Policy Assistant can be extended to the semantic analysis of privacy and EGovernance systems. Realistic examples are given throughout.