Hackers Stand to Win Nearly $20 Million in AI Cybersecurity Competition

The Biden Administration has unveiled a fresh hacking challenge, offering a lucrative prize to encourage the utilization of Artificial Intelligence (AI) in safeguarding critical U.S. infrastructure against cyber risks.
The AI Cyber Challenge promises nearly $20 million in prizes and will feature collaboration from leading AI companies including Anthropic, Google, Microsoft, and OpenAI, contributing their technology to the competition.
"We need to stay one step ahead in our defence. And artificial intelligence offers a very promising approach for that," Perri Adams, a program manager at DARPA's Information Innovation Office, remarked during a Tuesday telephone call with journalists.
"It's an opportunity to explore what's possible when cyber and AI experts have access to an unprecedented level of intercompany resources."
The preliminary event is set for spring, where up to 20 teams demonstrating the best results will be selected to advance to the semi-finals at DEF CON 2024, a cybersecurity conference.
Out of these, up to five teams will proceed to the DEF CON 2025 finals, each winning two million dollars.
The top three teams will also be eligible for additional prizes, with a grand prize of four million dollars for the team that "best secures critical life-saving software."
Competitors will be asked to open-source their system code so that their solutions can be widely employed.
Open Source Security Foundation at the Linux Foundation is also advising on this competition.
The contest's organizer, the Defence Advanced Research Projects Agency (DARPA), stated it will provide up to one million dollars to seven small businesses that wish to participate, aiming to diversify the pool of participants.
The government hopes that AI will further fortify the defence of the U.S.'s critically important systems.
This is not the first time the U.S. government has used hacking competitions to drive innovation. In 2014, DARPA initiated the Cyber Grand Challenge to develop an open-source automated defence system capable of shielding computers from cyberattacks, with a structure similar to this new challenge.