This study presents an advanced optimization model for the secure outsourcing of classified defense industrial research projects, with a particular focus on overcoming barriers to leveraging external organizational capacities. The core objective is to propose a rigorous solution that maximizes the effective use of external resources while safeguarding sensitive information. Information classification is recognized as a critical determinant of data security and confidentiality. Nevertheless, stringent classification protocols may impede information flow, reduce project efficiency, and complicate collaboration with scientific and industrial experts.To address these challenges, the research adopts a three-tiered hybrid methodology: (1) the Analytic Network Process (ANP) for comprehensive decision analysis, (2) Failure Mode and Effects Analysis (FMEA) for systematically identifying and managing confidentiality risks, and (3) mathematical allocation modeling to ensure secure and efficient outsourcing. The findings demonstrate marked improvements in both the preservation of security and the enhancement of operational efficiency. Among seven primary confidentiality criteria in defense projects, "macro-level security" (weight: 0.33) and "critical elements" (weight: 0.19) emerged as the highest priorities. The principal contribution of this work is the development of a multi-objective allocation model that rigorously balances outsourcing costs with confidentiality risks, thereby enabling optimized use of external capacities without compromising security. This framework not only ensures the protection of sensitive information but also promotes broader collaboration with national industrial and academic experts, advancing the interests of national defense. Ultimately, the proposed model provides policymakers with a robust, practical decision-making tool for secure outsourcing that aligns operational efficiency with stringent security imperatives.