In open and distributed systems, agents must engage in interactions
of which they have no previous experience. Deontic models are
widely
used to describe aspects of permission, obligation, and trust
anticipated by such agents, but no practical mechanism has been
developed for testing deontic trust specifications against models of
multi-agent interactions. In this talk we show how this is made
possible using a dynamic model checker. Our mechanism allows the
agents involved to dynamically and automatically invoke the model
checker at run-time to verify that certain trust rules will not be
broken.