A mechanism that allows enterprise authorities to be informed when security-sensitive decisions or actions have been or are attempting to be made by users of untrusted code executing in the trusted computing base. The mechanism may be implemented as an abstract class that is part of the trusted computing base. The class provides a framework abstract enough to permit multiple possible notifications (e.g., providing an e-mail to a system operator, sending an Simple Network Management Protocol (SNMP) alert, making an entry in an online database, or the like) in the event that a given action is taken by a user of untrusted code. The abstract class may provide a default notification, or the class may be extended to enable an authority to provide its own set of customized notifications.