The Biden administration on Wednesday launched an open prize competition aimed at generating artificial intelligence-based solutions to automatically secure the nation’s most critical software.

The AI Cyber Challenge (AIxCC) will be led by the Defense Advanced Research Projects Agency (DARPA) and include partners Anthropic, Google [GOOG], Microsoft [MSFT], and OpenAI, who will provide their expertise and technology for the competition.

The software most in need of protection is open-source and is used by critical infrastructure entities and software supply chains.

“AIxCC represents a first-of-its-kind collaboration between top AI companies, led by DARPA, to create AI-driven systems to help address one of society’s greatest challenges, cybersecurity,” Perri Adams, the agency’s program manager for the competition, said in a statement. “In the past decade, we’ve seen the development of promising new AI-enabled capabilities. When used responsibly, we see significant potential for this technology to be applied to key cybersecurity issues. By automatically defending critical software at scale, we can have the greatest impact for cybersecurity across the country, and the world.”

The competition is part of the administration’s ongoing commitment to the safe and responsible development of AI to solve the most pressing cybersecurity challenges.

AIxCC will proceed along two tracks, one funded and the other open. Competitors in the Funded Track will submit proposals in September in response to a Small Business Innovation Research solicitation while Open Track competitors will register through DARPA’s competition website to submit their proposals.

Up to 20 teams from both tracks combined will be invited to participate in the semifinals and of these the top five will receive $2 million each and continue to the finals. Additional funding will be awarded to the top three teams with first place winning $4 million, second place $3 million, and third place $1.5 million.

Launch of AIxCC was made at this week’s Black Hat information security conference. The semifinal winners will be announced in August 2024 at the annual DEF CON hacker convention and the final winners in August 2025 at DEF CON.

“If successful, AIxCC will not only produce the next generation of cybersecurity tools, but will show how AI can be used to better society by defending its critical underpinnings,” DARPA’s Adams said.