DARPA, the Defense Advanced Research Projects Agency, has announced a new competition that aims to design AI-based tools capable of detecting and resolving bugs in commonly used software. In collaboration with four top AI companies—Anthropic PBC, Microsoft Corp., Google LLC, and OpenAI Inc.—DARPA will launch this program, offering a total of $20 million in prize money, including a top prize of $4 million.
The annual DEFCON hacking conference, starting today in Las Vegas, will serve as the platform for the launch of this competition. Perri Adams, the program manager at DARPA overseeing the contest, revealed that the winning team will need to build a system that can rapidly defend critical infrastructure code from attack. This critical infrastructure code is used across various sectors, including chemicals, water, transportation, and more. The aim is to create an automated defense system that can safeguard software from potential attacks, ranging from commercial industry software to life-saving medical devices.
To ensure broad participation, DARPA will also provide funding to seven small businesses, each with $1 million, to help them compete in the initial phase of the competition. The semi-finals will be held at the 2024 DEFCON event, with finalists being selected during the 2025 event.
DARPA has previously used challenges and contests to advance various technologies, such as self-driving vehicles and underwater drones. Many of the technologies that emerged from these competitions have found commercial applications and have also been incorporated into the design of military systems.
This latest competition aims to bolster cybersecurity efforts by employing AI systems to identify and rectify security vulnerabilities in software before cybercriminals or spies from adversarial nations have a chance to exploit them. Anne Neuberger, the White House deputy national security adviser responsible for cybersecurity, highlighted the significance of this challenge, stating that it brings together cutting-edge automatic software security and AI capabilities, which will empower cyber defenses by quickly pinpointing and addressing software vulnerabilities.
While criminals and spies are also leveraging AI tools to expedite the discovery of vulnerabilities and accelerate the development of malware, the goal of the DARPA contest is to encourage a larger community of defenders to utilize AI in defense strategies. With an emphasis on AI-powered defense, this competition seeks to proactively combat cybersecurity threats and foster innovation in the field, ultimately leading to safer and more secure software systems.
The introduction of this competition aligns with the increasing importance of AI in bolstering cybersecurity measures. As technology evolves, so do the tactics employed by those seeking to exploit it. By harnessing AI to detect and respond to vulnerabilities, organizations can stay one step ahead of cyber threats and ensure the integrity and security of critical software systems. This initiative demonstrates the commitment of DARPA and its collaborators to fortify cybersecurity defenses and protect critical infrastructure.
