Title: Towards Language-Based Network Anonymity Authors: Aslan Askarov and Stephen Chong, Harvard University Abstract: Traditional systems for network anonymity are designed to be application agnostic. While this enables relatively simple deployment, many applications remain unaware of the anonymous nature of the underlying communication. Because security properties are usually application-specific, there is an opportunity to improve reasoning about anonymity guarantees by making applications aware of the anonymous nature of the underlying communication. This talk reports on our work in progress on a language-based approach to network anonymity. We distinguish between direct (or identifiable) and anonymous communication at the program source level. We introduce several classes of adversaries based on their ability to inspect anonymous traffic. A security type system regulates how anonymous information propagates within a program. This allows mixing of anonymous and identifiable communication within a single program, and may improve the overall performance while preserving anonymity.