{"id":4904,"date":"2024-08-15T17:25:24","date_gmt":"2024-08-15T22:25:24","guid":{"rendered":"https:\/\/www.darkreading.com\/application-security\/darpa-announces-ai-cyber-challenge-finalists"},"modified":"2024-08-15T17:25:24","modified_gmt":"2024-08-15T22:25:24","slug":"darpa-announces-ai-cyber-challenge-finalists","status":"publish","type":"post","link":"https:\/\/ddi.mohflo.net\/index.php\/2024\/08\/15\/darpa-announces-ai-cyber-challenge-finalists\/","title":{"rendered":"DARPA Announces AI Cyber Challenge Finalists"},"content":{"rendered":"
<\/a><\/div>\n
<\/div>\n

Seven teams advanced to the finals after the <\/span>DARPA AI Cyber Challenge<\/a><\/span> (AIxCC) semifinal competition was held at DEF CON 32 last week. The advancing teams were each awarded a $2 million prize and will participate in the final competition in August 2025. <\/span><\/p>\n

A partnership between DARPA and the Advanced Research Projects Agency for Health (ARPA-H), the <\/span>AIxCC<\/a><\/span> competition had teams design AI systems to secure open-source infrastructure software\u2013used across the private and public sectors in areas like finance, utilities and health care. These systems are vulnerable to cyberattacks as they have large attack surfaces and lack security tools at scale, and many have been recent targets, highlighting the urgent need to protect critical infrastructure. <\/span><\/p>\n

For the competition, teams were asked to develop Cyber Reasoning Systems that could find and fix vulnerabilities in a set of Challenge Projects that were designed by AIxCC experts. Nearly 40 teams submitted systems. The competitors\u2019 systems discovered 22 unique synthetic vulnerabilities in the Challenge Projects and patched 15 of them. The systems found 11 unique patches for C-based challenges and 4 for Java-based challenges. One real-world bug was found in SQLite3, and has been disclosed. <\/span><\/p>\n

\u201cIn true DARPA fashion, we didn\u2019t know if our hypothesis would be proven when we launched this program. Now, we\u2019ve seen that AI systems are capable of not only identifying but also patching vulnerabilities to safeguard the code that underpins critical infrastructure,\u201d said Andrew Carney, program manager for AIxCC in a statement.<\/span><\/p>\n

The following teams will advance to the final: <\/span><\/p>\n

\n