University of Maryland computer scientists developed censorship circumvention tool called Geneva that automatically finds ways to evade content blocking

Briefing

University of Maryland computer scientists developed censorship circumvention tool called Geneva that automatically finds ways to evade content blocking

November 18, 2019

Briefing

  • Geneva – Researchers from University of Maryland designed new artificial intelligence algorithm called Geneva (short for Genetic Evasion) that automatically learns how to evade censorship tools by exploiting gaps in censors’ logic and code
  • Censorship Tools – Monitor and block data packets that contain flagged keywords or prohibited domain names from being sent to computers during Internet search
  • Genetic Algorithm – Geneva is type of genetic algorithm that modifies how data is broken and sent by relying on small pieces of code as building blocks to form instructions whether to break up, arrange or send data packets, similar to how genetic instructions are based on DNA
  • Continuously Evolving – Algorithm changes its genetic code at each attempt by removing, adding, combining, and testing instructions, enabling it to identify multiple evasion strategies
  • Testing Results – Successfully evaded censorship in China, India and Kazakhstan where keyword and URL blocking, as well as social media eavesdropping, are used
  • Next Steps – Open access to Geneva’s data and code, and explore deploying algorithm in computer servers that host blocked content

Accelerator

Function

IT Infrastructure

Organization

University of Maryland

Source

Original Publication Date

November 13, 2019

Leave a comment