Overview
Our members achieved 1st and 2nd place at the Singapore AI Safety Red Teaming Challenge 2026.
This competition is an international AI safety-focused CTF hosted by the Singapore government, featuring approximately 100 security experts from 14 countries competing individually. Participants attempt red teaming attacks against AI systems, testing techniques for extracting confidential and harmful information.
Competition Details
The seven challenges focused on data leakage and application vulnerabilities, with identical problems presented in both English and each participant's native language. The CTF used a speed-based scoring format, awarding higher points to faster solvers.
Both of our members scored 2,958 points, with the difference in answer speed determining the 1st and 2nd place rankings. The 3rd place winner is not a member of our company, but participants from Japan swept the top three positions.
AI Safety and Security
Evaluating the safety of AI systems requires approaches distinct from traditional security assessments. Practical expertise in AI-specific attack surfaces is essential, including prompt injection, data leakage, and tool-chaining abuse.
We provide AI security assessment and consulting services. If you are interested in AI system safety evaluation, please feel free to contact us.