A User-Friendly Approach to Human Authentication of Messages

Users are often forced to trust potentially malicious terminals when trying to interact with a remote secure system. This paper presents an approach for ensuring the integrity and authenticity of messages sent through an untrusted terminal by a user to a remote trusted computing base and vice versa. The approach is both secure and easy to use. It leverages the difficulty computers have in addressing some artificial intelligence problems and therefore requires no complex computation on the part of the user. This paper describes the general form of the approach, analyzes its security and user-friendliness, and describes an example implementation based on rendering a 3-D scene.