In a collaborative effort backed by South Korea's Ministry of Science and ICT (MSIT), the AI startup VIDRAFT and Naver Cloud have joined a government-funded initiative to develop security-specialized AI models. Unlike general-purpose large language models, this project focuses on creating foundation or fine-tuned models specifically optimized for cybersecurity use cases. A core technical component of this initiative is the application of VIDRAFT's proprietary AX-RAY technology, which specializes in deep document understanding and information extraction. This technology is intended to enhance the models' ability to perform structural and semantic analysis, which is critical for security-centric tasks such as automated vulnerability analysis, parsing policy documents, and summarizing threat intelligence reports. By integrating AX-RAY's capabilities with security-domain-specific datasets—such as malware signatures and vulnerability reports—the collaboration aims to build architectures that utilize retrieval-augmented generation (RAG) or structured extraction to reduce hallucination in security-critical environments. Naver Cloud is expected to provide the necessary cloud-native infrastructure for these models. While specific quantitative benchmarks and model architectures remain undisclosed, the partnership represents a significant public-private effort to transition high-level document reasoning into production-grade cybersecurity tooling under the MSIT framework.