US Government Launches AI Cyber Challenge to Protect Critical Infrastructure

The administration of Joe Biden and Kamala Harris launched a contest open to small American technology businesses that want to create an artificial intelligence system that prevents attacks on government software.

It was announced that the project has the collaboration of Anthropic, Google, Microsoft and OpenAI.

The Artificial Intelligence Cyber ​​Challenge (AIxCC) is organized and led by the United States Defense Advanced Research Projects Agency (DARPA).

“This competition, which will have nearly 20 million dollars in prizes, will promote the creation of new technologies to rapidly improve the security of computer code, one of the most pressing challenges in cybersecurity,” said the White House.

To ensure broad participation and a level playing field for AIxCC, DARPA will also make $7 million available to small businesses that want to compete. The contest has a total duration of two years and these are its phases:

1. Announcement

According to the AIxCC page, the registration of interested companies is carried out during the fall, from November 1 to December 15, 2023, through a form provided by Darpa. At this point, seven small businesses are chosen to receive seed funding.

2. Semifinals

In August 2024 there will be the semifinals, there will be only five teams that will receive 2 million dollars each for the timely development of their projects. Not before going through some preliminary and pre-qualifying events in the spring of next year.

3. Finals

In August 2025, three winners will be announced who will obtain a different amount of money according to their final positions. First place will have 4 million dollars; the second, 3 million and the third, 1.5 million.

Major competitors will play a crucial role in shaping the cybersecurity landscape for both the United States and the global community. The Linux Foundation’s Open Source Security Foundation is also acting as an advisor on this challenge.

“Our objective in the DARPA AI challenge is to stimulate the growth of a wider community of cyber defenders by leveraging the participating AI models to enhance the speed and effectiveness of our cyber defenses. Our aim is to harness the power of generative AI to strengthen and fortify our cybersecurity measures,” said Anne Neuberger, Deputy Homeland Security Advisor for Technologies Cybernetics and Emerging States of the United States.

This competition was showcased at the renowned Black Hat USA Conference held in Las Vegas, which is recognized as the largest hacking event in North America. This is not the first time that the US government has announced a contest at such a meeting, in 2014, Darpa launched the Cyber ​​Grand Challenge to develop an open source automatic defense system that could protect a computer from cyber attacks, with a structure similar to new challenge of two years.

One of the most recent cyber attacks in the United States was precisely to its airports. These sites fell victim to a hack organized by a pro-Russian hacker group known as KillNet in October last year.

In total, just over a dozen airports were affected, according to a CNN count, including Hartsfield-Jackson in Atlanta and Los Angeles, whose website was operating very slowly.

Parts of the Los Angeles International Airport website were also disrupted, spokeswoman Victoria Spilabotte said. “There were no compromises to the internal airport systems, and no disruptions to operations were reported.”

Spilabotte mentioned that the airport informed both the FBI and the Transportation Security Administration.