Hello, Guest!


DARPA Launches Challenge to Develop AI-Powered Vulnerability Detection Tech

Cyber reasoning system

DARPA Launches Challenge to Develop AI-Powered Vulnerability Detection Tech

The Defense Advanced Research Projects Agency is accepting applications for the AI Cyber Challenge, a competition to develop artificial intelligence-powered cyber reasoning systems that detect and remediate software vulnerabilities in real time and at scale.

Winning teams stand to receive up to $29.5 million in cumulative prize money throughout the program, DARPA said.

AIxCC will have a track exclusively for small businesses, replacing a previously announced funded track. DARPA said it would award seven small businesses $1 million each and is encouraging organizations that applied to the funded track to reregister with the small business one.

Submissions for the small business track are due Jan. 15, 2024. Meanwhile, registration for AIxCC’s open track will last until April 30, 2024.

According to DARPA, the competition’s grading system will favor CRS projects that cover a diverse array of vulnerabilities, “accurately identify scorable vulnerabilities” and execute effective fixes without impacting software functionality. The system is expected to penalize projects that spam inaccurate patches.

The Open Source Security Foundation will contribute to AIxCC’s challenge design, while DARPA partners Anthropic, Google, Microsoft and OpenAI will provide access to credits for their large language models.

Sign Up Now! Potomac Officers Club provides you with Daily Updates and News Briefings about Cybersecurity

Category: Cybersecurity

Tags: AI Cyber Challenge artificial intelligence cyber reasoning system cybersecurity Defense Advanced Research Projects Agency software vulnerability