The Korea Advanced Institute of Science and Technology (KAIST, President Kwang-Hyung Lee) and Samsung Electronics (CEO Jong-Hee Han) announced on the twenty first that their joint team, ‘Team Atlanta,’ has been chosen because the finalist for the AI Cyber Challenge to be held in August next 12 months.
This competition was held from the eighth to the eleventh at DEFCON, one in every of the highest cybersecurity conferences in Las Vegas, USA, led by the Defense Advanced Research Projects Agency (DARPA), a research institute under the U.S. Department of Defense. It’s a contest for the next-generation hacking system using AI.
Team Atlanta is a joint team of ▲KAIST ▲Samsung Research ▲POSTECH ▲Georgia Tech, and is principally composed of members from the lab of Professor Taesoo Kim of Georgia Tech, who’s currently serving because the Executive Director of Samsung Research.
Team Atlanta placed in the highest seven on this qualifier and received $2 million in research funding, advancing to the finals at Defcon next August.
Professor Insoo Yoon of KAIST is a world-class white hacker and professor who has won ‘Defcon CTF’ twice and won an award at ‘Pwn2Own 2020’, an American hacking competition. He has repeatedly presented his research at the very best academic societies in the sphere of security and has received the very best paper award at international academic conferences.
This AI Cyber Challenge is a contest for AI-based cyber reasoning systems (CRS). DARPA posed problems including past vulnerabilities or artificial vulnerabilities to real software reminiscent of Linux, and every team’s CRS robotically analyzed the software to discover and patch vulnerabilities. Afterwards, each CRS was evaluated comprehensively considering the number and number of vulnerabilities discovered, the accuracy of patches, etc.
Particularly, Team Atlanta’s CRS discovered a brand new vulnerability in SQLite3, a famous software that was included within the preliminary round, that was not intended by the examiner. That is evaluated as a very important achievement that shows the potential for innovation that AI can bring to the safety field and can also be in keeping with the goal of the AI Cyber Challenge.
Professor Yoon In-soo said, “I consider that this result shows that the safety research capabilities of Korea, including KAIST, have reached a world-class level,” and added, “Over the subsequent 12 months, we are going to try more groundbreaking and proactive methods, lead innovation within the convergence of AI and security, and do our greatest to win this competition.”
Reporter Park Soo-bin sbin08@aitimes.com
