DARPA to Bring AI Cyber Challenge Semifinal Competition to DEF CON 32

AIxCC Experience Conceptual Art (DARPA)

From August 9-11, in Las Vegas, the Defense Advanced Research Projects Agency (DARPA) in collaboration with the Advanced Research Projects Agency for Health (ARPA-H) will hold the AI Cyber Challenge (AIxCC) Semifinal Competition event at DEF CON 32 - one of the world's largest cybersecurity conferences - at the Las Vegas Convention Center.

AIxCC is an ambitious two-year competition that brings together the foremost experts in AI and cybersecurity to create novel AI systems that can safeguard the open-source software critical to modern life. The AIxCC Semifinal Competition will unfold at DEF CON in a large-scale, immersive and interactive experience that brings to life the challenges of securing critical infrastructure in the context of a futuristic city environment. The experience invites attendees to explore the possibilities of AI-driven cybersecurity solutions while visualizing the competing teams' progress.

"Critical infrastructure software - which undergirds everything from financial systems and hospitals to public utilities - is pervasively vulnerable to cyber attacks¹ . These attacks can be carried out remotely, causing significant physical as well as digital damage. Through AIxCC, we're exploring whether AI can help find and fix software vulnerabilities more effectively and usher in a future where we can stop attacks as fast as we can detect them," said Andrew Carney, program manager for AIxCC. "Our goal is to inspire anyone who can advance AI-driven cybersecurity to get involved with efforts to secure the systems we all depend on."

Registration to compete in AIxCC closed in April 2024. For the Semifinal Competition, teams were tasked with developing a Cyber Reasoning System that will be tested against a set of Challenge Projects - identical for each team - that have been designed and evaluated by subject matter experts. Each Challenge Project has a basis in a real-world, open-source project that is critical to industry, national security, and the public, and contains vulnerabilities for teams' cyber reasoning systems to identify and secure. Teams will be scored according to an algorithm based on their ability to find and fix vulnerabilities. AIxCC received nearly 40 Cyber Reasoning System submissions for the Semifinal Competition.

At the event, large-screen visualizations will show updates on the competing teams' progress, including how teams are performing in relation to each other, how they are progressing in identifying each of the vulnerabilities within the competition's Challenge Projects, and successful patches. Fictional scenarios will translate the largely cloud-based competition to the real world with a wide range of content for attendees to explore and learn from.

"Throughout DEF CON, we'll share updates that help to answer the question: is this working? Does AI-driven cybersecurity show promise for securing increasingly connected and at-risk critical infrastructure? We look forward to sharing initial results with the community, both at DEF CON and in the months that follow," said Carney.

The AIxCC stage will feature programming from AI and cybersecurity experts throughout the event. In addition, DARPA leadership will speak on the DEF CON main stage, and deliver a keynote at the preceding Black Hat AI Summit, also in Las Vegas. Upcoming talks and activities include the following:

• On Tuesday, Aug. 6, at 11:40 a.m. PT at the Mandalay Bay, DARPA Information Innovation Office Director Kathleen Fisher will present "Enhancing National Security with AI-driven Cybersecurity" at the Black Hat AI Summit.
• On Friday, Aug. 9, at 1:00 p.m. PT, DARPA Director Stefanie Tompkins will deliver a talk titled, "If Existing Cyber Vulnerabilities Magically Disappeared Overnight, What Would Be Next?" and will be joined on stage by ARPA-H Director Renee Wegrzyn and DARPA Chief Information Officer Peiter "Mudge" Zatko for discussion.
• On Sunday, Aug. 11, at 12:30 p.m. PT, DARPA AIxCC Program Manager Andrew Carney and DARPA Special Assistant to the Director Perri Adams will discuss initial results from the AIxCC Semifinal Competition.
• In the DEF CON AI Village, DARPA's Semantic Forensics team will demo technologies that can help people defend against threats posed by deepfakes.

Those not attending DEF CON in person will be able to view AIxCC Semifinal Competition content and updates at https://aicyberchallenge.com/.

AIxCC follows the DARPA Challenge model, which applies the power of competition as a forcing function to drive innovation. DARPA is working with industry leaders for this first-of-its-kind collaboration. Anthropic, Google, Microsoft, and OpenAI are providing their subject matter expertise and making their technology available to competitors. The Linux Foundation's Open Source Security Foundation is a challenge advisor to ensure that resulting technology can integrate into the open-source community to secure critical, open-source software. For more information, please visit https://aicyberchallenge.com/.

Media Opportunities

The AIxCC Semifinal Competition experience at DEF CON will be open to press who are registered to attend DEF CON 32. Media are also welcome to schedule time to leverage the AIxCC Studio on site, which will offer a dedicated AIxCC interview space, broadcast-grade cameras with broadcast-quality recording capability, an XR-enabled, virtualized environment with engaging and interactive visuals, broadcast-quality lighting, and a studio crew. Media interested in arranging interviews with AIxCC experts or in scheduling time in the Studio should email [email protected].

[1] https://www.fbi.gov/news/stories/chinese-government-poses-broad-and-unrelenting-threat-to-u-s-critical-infrastructure-fbi-director-says