Skip Navigation

Produced by the Office of Marketing and Communications

Subscribe Now

A Leg Up in the Censorship Arms Race

New Artificial Intelligence System Automatically Evolves to Evade Regimes’ Internet Controls

By Kimbra Cutlip

GIF of "Access Denied" message flashing on a laptop screen as hands type

Animation by Dillon Shook/Unsplash

New work led by UMD computer scientists could shift the balance of power in the censorship race, using a new artificial intelligence tool inspired by the principles of genetic evolution.

Internet censorship by authoritarian governments handcuffs millions of web users daily, and researchers scrambling to circumvent it find themselves in a continually escalating race to keep up with ever-changing, increasingly sophisticated controls.

New work led by University of Maryland computer scientists could shift the balance of power in the censorship race, using a new artificial intelligence tool inspired by the principles of genetic evolution. They named it Geneva (short for Genetic Evasion), and rolled it out on Thursday in a peer-reviewed talk at the Association for Computing Machinery’s 26th Conference on Computer and Communications Security in London.

Geneva automatically learns how to circumvent censorship. In tests in China, India and Kazakhstan, it found dozens of ways to beat censorship by exploiting gaps in censors’ logic and finding bugs that the researchers say would have been virtually impossible for humans to find manually.

“Geneva represents the first step toward a whole new arms race in which artificial intelligence systems of censors and evaders compete with one another,” said Dave Levin, an assistant professor of computer science at UMD and senior author of the paper. “Ultimately, winning this race means bringing free speech and open communication to millions of users around the world who currently don’t have them.”

Information on the internet is broken into data packets by the sender’s computer and reassembled by the receiving computer; a prevalent form of censorship used by authoritarian regimes works by monitoring the data packets sent during an internet search and blocking requests that either contain flagged keywords or domain names. But when Geneva is running on a computer that is sending out web requests through a censor, it modifies how data is broken up and sent, rendering the censor unable recognize forbidden content or censor the connection.

Known as a genetic algorithm, Geneva is a biologically inspired type of artificial intelligence that Levin and his team developed to work in the background as a user browses the web from a standard browser. Like biological systems, Geneva forms sets of instructions from genetic building blocks. But rather than DNA, Geneva uses small pieces of computer code as those building blocks. Individually, the bits of code do very little, but when composed into instructions, they can execute sophisticated evasion strategies for breaking up, arranging or sending data packets.

Geneva evolves its genetic code through successive attempts, or “generations.” With each generation, Geneva keeps the instructions that work best at evading censorship and kicks out the rest. Through the evolutionary process, Geneva is able to identify multiple evasion strategies very quickly, inverting how researchers typically approach the problem of censorship.

“Ordinarily we identify how a censorship strategy works and then devise strategies to evade it,” said Levin, who holds a joint appointment in the University of Maryland Institute for Advanced Computer Studies. But now we let Geneva figure out how to evade the censor, and then we learn what censorship strategies are being used by seeing how Geneva defeated them.”

After lab tests, the team demonstrated that Geneva worked in the real world against undiscovered censorship methods. By deploying strategies identified by Geneva, a computer user in China was able to browse free of keyword censorship. The researchers also successfully evaded censorship in India, which blocks forbidden URLs, and Kazakhstan, which was eavesdropping on certain social media sites at the time. In all cases, Geneva successfully circumvented censorship.

“Currently, the evade-detect cycle requires extensive manual measurement, reverse engineering and creativity to develop new means of censorship evasion,” said Kevin Bock ’17, M.S. ’18, a computer science Ph.D. student at UMD and lead author of the paper. “With this research, Geneva represents an important first step in automating censorship evasion.”

The researchers plan to release their data and code in the hopes that it will provide open access to information in countries where the internet is restricted.

“If Geneva can be deployed on the server side and work as well as it does on the client side, then it could potentially open up communications for millions of people,” Levin said. “That’s an amazing possibility, and it’s a direction we’re pursuing.”




Maryland Today is produced by the Office of Marketing and Communications for the University of Maryland community on weekdays during the academic year, except for university holidays.