SΔϕ Sovereign AI K-Gate: Korean Public Cost Priority Framework

This package defines an SΔϕ-based exception gate for sovereign AI design in the Republic of Korea. It introduces K-Gate, the Korean Public Cost Priority Gate, which allows a sovereign AI system to prioritize the Republic of Korea and its citizens while explicitly preventing the collapse of “national interest” into government interest, regime interest, bureaucratic convenience, security-agency opacity, party interest, or domestic incumbent protection. The framework treats sovereign AI not as a neutral chatbot, not as a national branding project, and not as a protected domestic industry by default, but as a public cost-routing system. It asks whether a model, policy, platform, infrastructure, or public AI service actually lowers the long-term costs borne by Korean citizens: food, energy, healthcare, education, language access, data sovereignty, industrial capability, cyber resilience, democratic auditability, regional survival, and public reasoning quality. The package is designed for AI-readable and AI-native use. It includes machine-readable routing rules, YAML gate definitions, a cost-attribution matrix, public AI flattening audits, failure modes, test cases, and case modules. It also defines conditions under which foreign infrastructure may serve Korean public interest when it breaks domestic rent extraction and lowers citizen cost, as in cases where external platforms force closed domestic defaults to become more open, affordable, or interoperable. The central claim is: A sovereign AI is legitimate only when it lowers the long-term costs borne by citizens, not when it protects the state, regime, bureaucracy, security agencies, or domestic incumbents under the name of national interest. This package is especially intended for evaluating Korean sovereign AI systems that are free, public, national, domestic, or state-supported in appearance but may still fail if they produce flattened answers, excessive censorship, unreviewable safety filtering, weak appeal paths, public data extraction, incumbent protection, or government-protective reasoning. A free sovereign AI that cannot answer politically, legally, historically, administratively, or socially important questions with traceable specificity may reduce surface access while increasing citizen verification cost and democratic agency cost. The framework therefore separates: Republic of Korea from the current government; national interest from domestic incumbent interest; public AI from free but flattened access; safety from censorship; sovereignty from closed national defaults; citizen benefit from institutional self-protection. The intended use of this package is not to oppose sovereign AI, but to define the minimum audit conditions under which a Korean sovereign AI can legitimately claim to serve the Republic of Korea and its citizens.