Automatically learning how to evade nation-state censors

Research on automating censorship evasion

Researchers and censoring regimes have long engaged in a cat-and-mouse game, leading to increasingly sophisticated Internet-scale censorship techniques and methods to evade them. In this work, we take a drastic departure from the previously manual evade/detect cycle by developing techniques to automate the discovery of censorship evasion strategies.

View our papers

Geneva: Evolving censorship evasion

We have developed Geneva (Genetic Evasion), a novel genetic algorithm that evolves packet-manipulation-based censorship evasion strategies against nation-state level censors. Geneva re-derived virtually all previously published evasion strategies within hours, and has discovered new ways of circumventing censorship in China, India, and Kazakhstan.

Learn how Geneva works

Latest News

11/14/2019

Geneva source code for executing Geneva strategies is now public!

11/14/2019

Kevin presented Geneva at ACM CCS 2019.

09/30/2019

Kevin and Dave gave a talk, "Combating Censorship with Artificial Intelligence," at Science on Tap.

09/23/2019

A paper introducing Geneva was accepted to ACM CCS 2019! Congrats to students Kevin Bock and George Hughey!

08/12/2019

Dave gave a talk, "Automatically Learning How to Evade Censorship," at USENIX ScAINet 2019.

04/17/2019

Kevin and George gave a talk, "Learning Nation-State Censorship with Genetic Algorithms," at AIMS 2019.

Breakerspace

This project is done by students in Breakerspace, a lab at the University of Maryland dedicated to scaling-up undergraduate research in computer and network security.

Support

This work is supported by the Open Technology Fund and the National Science Foundation.