SK Telecom has officially joined Project Glasswing, an initiative led by the AI safety and research company Anthropic, gaining early access to its powerful cyber-defense AI model, Claude Mythos.
This move places one of South Korea's largest telecommunication companies at the forefront of AI-driven cybersecurity. Project Glasswing is an exclusive consortium of major tech companies and government agencies, initially focused on the U.S., designed to use frontier AI to find and fix critical vulnerabilities in software that powers our daily lives—before malicious actors can exploit them. The AI at its core, Claude Mythos, has demonstrated capabilities so advanced that its access is being carefully restricted. For instance, the UK's AI Safety Institute (AISI) found it could autonomously complete a complex, 32-step cyberattack simulation, a milestone that highlights both its potential for defense and the risks of misuse.
So, why did this happen now? The decision is rooted in a clear causal chain. First, a major driver is the shifting global policy landscape. The U.S. government has been tightening controls on AI technology, such as through a recent executive order encouraging AI labs to share their most powerful models for national security vetting. This creates a strong incentive for allies like South Korea to join controlled, trusted programs like Glasswing to stay ahead of the curve.
Second, the urgency is amplified by a growing threat environment. Reports from 2025 showed a significant surge in cyber intrusions in Korea and record-breaking crypto theft by state-sponsored actors like those from North Korea. This creates immense pressure to upgrade national cyber-defenses, making a defensive tool like Mythos highly valuable. Finally, the opportunity arose as Project Glasswing began expanding its membership beyond the initial U.S. cohort to include other allied nations, opening a window for South Korean entities to participate.
This partnership is more than just a technical upgrade; it's a strategic alignment that strengthens the U.S.-South Korea technology alliance. For SK Telecom, it not only enhances its ability to protect its own critical network infrastructure but also opens the door to developing and selling next-generation, AI-assisted security services to other businesses. The market anticipated this move, with SKT's stock price significantly outperforming the market index just before the announcement, signaling strong investor confidence in the strategic value of this collaboration.
[Glossary]
- Project Glasswing: An initiative by the AI company Anthropic, bringing together key technology firms and government bodies to use advanced AI for identifying and fixing software vulnerabilities in critical infrastructure.
- Claude Mythos: A highly advanced 'frontier' AI model developed by Anthropic, specifically designed for complex problem-solving in areas like cybersecurity. Its capabilities are powerful enough to warrant restricted access.
- Frontier Model: A term for the most powerful and capable AI models currently in existence. These models push the boundaries of AI capabilities and often require special safety considerations.