South Korea aims to become one of the world’s three leading AI powers, but its strongest model ranked only 36th in an Artificial Analysis evaluation and scored 63% of the top model’s result. The global top 25 comprised 15 U.S. models and 10 Chinese models, led by Anthropic’s Claude Fable 5.1 and OpenAI’s GPT-6 Astra. South Korea also trails in research output and private investment: Stanford University’s AI Index 2026 counted 50 notable U.S. models, 30 Chinese models and five South Korean models last year, while private AI investment totaled $285.9 billion in the U.S. and $12.4 billion in China. As leading U.S. companies argue for safety-related deceleration, South Korean officials and industry executives warn that slowing development could restrict latecomers’ ability to catch up. Deputy Prime Minister and Minister of Science and ICT Bae Kyung-hoon said South Korea cannot afford to slow AI development, while stressing faster progress alongside safety verification, security and resource efficiency. The government is prioritizing indigenous foundation models, security-focused AI, physical AI and frontier capabilities, even as industry voices differ over how closely development speed should track risk-management capacity.