DARPA is set to evaluate seven teams’ cyber reasoning systems in a competition at the upcoming DEF CON conference in Las Vegas. The finalist teams will demonstrate their models' ability to autonomously identify and patch vulnerabilities in open-source code. These AI-powered systems aim to enhance security for critical infrastructure sectors that depend on open-source software. Open-source tools are beneficial for operators but also more vulnerable to attacks. The competition reflects advancements in large language models and aims to improve cybersecurity methods for public safety.
The Defense Advanced Research Projects Agency is evaluating seven competing teams' cyber reasoning systems at a competition during the DEF CON hacker conference in Las Vegas.
The top three teams will be chosen based on their ability to autonomously identify and patch vulnerabilities in open-source code.
Open-source tools are vulnerable to cyber exploitation due to publicly available code that allows attackers to easily identify and exploit weaknesses.
Over 7.8 million lines of code were analyzed by teams, which identified 59% of synthetic vulnerabilities and patched 43%.
Collection
[
|
...
]