A hacking defense competition will be held where participants use artificial intelligence (AI) to prevent cyberattacks and identify vulnerabilities in AI systems themselves. This year, a new challenge has been added that tests how effectively the latent capabilities of AI can be leveraged.
The Korea Internet & Security Agency (KISA) announced on the 8th that it is recruiting preliminary participants for the '2026 AI Hacking Defense Competition (ACDC)' in collaboration with the Ministry of Science and ICT and the Korea Information Protection Industry Association, with applications open until the 26th.
The ACDC, first held last year, marks its second edition this year. Amidst the sophistication of cyberattacks driven by advancements in AI technology, the competition is characterized by evaluating not only defensive capabilities using AI but also the ability to identify and respond to security vulnerabilities within AI itself.
The competition covers three core areas of AI security: 'Security Utilizing AI,' which involves detecting and responding to cyber threats with AI; 'Ensuring AI Safety,' which focuses on protecting AI models themselves from attacks; and 'AI Platform Security,' which aims to protect the systems and infrastructure on which AI operates.
The preliminary round, conducted online on the 31st, will feature problems reflecting real-world attack and defense scenarios. Each participating team will be provided with an isolated cloud environment and must solve security issues within a time limit.
There are two competition formats. 'Capture The Flag (CTF)' involves analyzing system vulnerabilities or solving problems to find hidden specific strings. In 'King of the Hill,' teams occupy a given system and earn higher scores by maintaining their occupation for longer periods while defending against attacks from other teams.
Notably, a new challenge not present last year will be introduced in this year's finals. This task evaluates how effectively the potential performance and capabilities of AI can be utilized, with specific competition formats and details to be disclosed later.
The top 40 teams from the preliminary round will advance to the finals. The finalists consist of a total of 160 people, divided into 20 teams in the general division and 20 teams in the student division. The finals will be held on December 1 at the Seoul Grand InterContinental Hotel Parana in Seoul. Outstanding teams will receive awards including the Vice Prime Minister and Minister of Science and ICT Award, the KISA Director General Award, and special prizes, along with prize money and gifts.
Eligibility is open to Korean nationals aged 19 and older. Anyone interested in AI and cybersecurity can form a team of four to apply. The application deadline is at 2 p.m. on the 26th of this month.
Oh Jin-young, Head of the AI Security Industry Division at KISA, stated, "As AI technology rapidly evolves, we must prepare not only for attack automation utilizing AI but also for new security threats targeting AI itself and the environment in which it operates." He added, "We expect this competition to serve as an opportunity to discover practical talent and enhance our country's AI security capabilities."
in TECH
