The mobile agents paradigm integrates a network of computers in a novel way and, in a certain sense, reduces networking to program construction. A mobile agent can travel from one place to another and, subject to the destination's approval, interact programmatically with the place it visits. Besides a good deal of promise, mobile agents also introduce new problems. Most of these problems have to do with security and safety. Some of the security concerns have been studied by the distributed systems community and cryptologists for a long time, but mechanisms and technologies developed to secure communication and control access to resources must be adapted to take into account mobility and protection of mobile agents from malicious hosts.