US launches DARPA AI cyber challenge: Prioritises cyber defence through AI software security
Two-year contest culminating at DEF CON 2025, winners may open-source their AI systems. Advancing cybersecurity and harnessing AI for critical digital protection.
Highlights
- DARPA's AI Cyber Challenge unfolds over two years, emphasising a long-term commitment to digital security enhancement
- Winners may opt to open source AI systems, promoting knowledge sharing and contributing to AI advancement
- DARPA will make available $1 million each to up to seven small businesses keen on participating
- A substantial $18.5 million in prizes will be awarded to top performers
In a concerted effort to bolster cybersecurity, the Defense Advanced Research Projects Agency (DARPA) in the US has announced an upcoming two-year contest called the AI Cyber Challenge. This initiative is part of the ongoing White House campaign to enhance software security, and it's not just the United States that's taking these steps – other countries are also championing safe AI and web practices through similar competitions.
DARPA's challenge will task participants with utilising AI to identify and rectify software vulnerabilities, specifically focusing on safeguarding critical infrastructure code.
Advancing cyber defence with AI: DARPA's innovative contest to safeguard critical software
Partnering with key players such as Microsoft, Google, and AI startups Anthropic and OpenAI, the competition will engage U.S.-based teams in a bid to secure essential software using AI techniques.
The Linux Foundation's Open Source Security Foundation (OpenSSF) will serve as an advisory partner, and a substantial $18.5 million in prizes will be awarded to top performers. Furthermore, DARPA will make available $1 million each to up to seven small businesses keen on participating.
The contest aims to create automated defence systems capable of safeguarding various software from potential cyberattacks. This becomes increasingly vital as open-source code gains prominence in critical software applications.
A recent GitHub survey revealed that a staggering 97 percent of apps leverage open-source code, with 90 percent of companies incorporating it in some capacity. However, the surge in open-source usage has also led to new vulnerabilities and exploits.
In light of damaging incidents like the Colonial Pipeline ransomware attack and SolarWinds supply chain breach, the Biden-Harris Administration has taken measures to enhance software supply chain security, including creating a cybersecurity safety review board.
Now, with the AI Cyber Challenge, the administration is placing its faith in AI's potential to significantly contribute to cyber defence efforts. The contest will unfold over two years, encompassing qualifying events and culminating in a final phase at DEF CON 2025.
Winners will have the option, though not an obligation, to open source their AI systems. This challenge stands as a pivotal endeavour in not only shoring up cybersecurity but also exploring how AI can be harnessed for the greater good in safeguarding critical digital foundations.
