DARPA, White House launch $20M AI, cybersecurity challenge
The Pentagon logo, after some technical difficulties. (Graphic by Breaking Defense)
WASHINGTON — The Defense Advanced Research Projects Agency (DARPA) and the White House have teamed up to announce a new two-year competition to use artificial intelligence to help protect critical software and create new cybersecurity tools in what’s essentially a cutting-edge, high-stakes bug bounty program.
Dubbed the “AI Cyber Challenge,” or AIxCC, the effort aims to “challenge competitors across the United States, to identify and fix software vulnerabilities using AI,” the White House announced today. Google, Microsoft, OpenAI and Anthropic will lend expertise and technologies for the challenge.
“In an increasingly interconnected world, software undergirds everything from financial systems to public utilities,” according to a DARPA press release. “As software enables modern life and drives productivity, it also creates an expanding attack surface for malicious actors.
“This surface includes critical infrastructure, which DARPA experts say is especially vulnerable to cyberattacks given the lack of tools capable of securing systems at scale,” the press release says. “Recent years have exposed the threats posed to society by malicious cyber actors exploiting this state of affairs, and have made plain the daunting attack surface cyber defenders are tasked to protect. Despite these vulnerabilities, advances in modern technology may provide a path towards solving them.”
RELATED: Pentagon memo aims to leverage $9B JWCC ‘to greatest extent possible’
DARPA will host an open competition for AIxCC where up to 20 teams will advance to the semifinals next August at the DEF CON 2024 conference, followed by up to five teams advancing to the finals, according to the agency’s website. In August 2025, three winners will be chosen at DEF CON 2025. AIxCC will feature almost $20 million in prizes, according to the White House.
AIxCC follows the Pentagon’s, and White House’s, broader push to use and adopt AI responsibly to advance national security. Last June, DoD released its Responsible AI Strategy and Implementation Pathway, carving out ways to transform the department into an AI-ready organization, and two years prior, the Pentagon adopted its AI Ethical Principles.
And just last month, the White House announced several leading AI companies, including those providing tech and expertise for AIxCC, had made a voluntary commitment to manage risks posed by AI.
Biden Takes Aim At US Investment In Chinese Tech
Meanwhile, the Biden administration also unveiled moves to guard against what it dubbed the “extraordinary threat” of rival China’s AI development by tightening the reins on American investment there. A new executive order, announced today along with a new Treasury program in the works, aims to prohibit investors from pouring their money into private Chinese tech firms who deal with AI, quantum computing and microelectronics that would benefit the Chinese military or intelligence services.
More than the money, senior administration officials told reporters the EO is meant to restrain the “intangible” benefits that often come with significant Western venture and private investment, including the introduction of new expertise, talent networks and “market access.”
“The United States already prohibits or restricts the export to the PRC [People’s Republic of China] of many of the technologies and products under consideration for the new program,” the Treasury Department said of its newest initiative. “This new program would prevent U.S. investments from helping accelerate the indigenization of these technologies in the PRC, which undermines the effectiveness of our existing export controls and inbound investment screening programs which also seek to protect U.S. national security.”
Breaking Defense’s Lee Ferran contributed to this report.