FIJISHI EPL

Japan's Defence Ministry issued its first official guidelines on responsible use of AI in defence research and development in June 2025 explicitly prohibiting lethal autonomous weapon systems and mandating human oversight of autonomous systems. Japan participated in AUKUS Pillar II maritime autonomous systems exercises in July 2025. Japan's Digital Japan 2026 policy explicitly distinguishes between Sovereign AI - infrastructure ownership - and AI Sovereignty - constitutional command authority over autonomous systems. The Defence Ministry's human oversight mandate requires that human control be exercised over autonomous decisions but no technical specification in Japan's defence AI guidelines defines what cryptographic proof of human control looks like or how it is produced for subsequent legal or parliamentary examination. 58 percent of organisations in Japan's defence environment have AI agents acting autonomously while 17 percent report limited or no human oversight. This paper documents the structural gap between Japan's human-centric defence AI mandate and the constitutional command architecture required to make human control technically verifiable rather than administratively asserted.

Read paper: https://doi.org/10.5281/zenodo.20739271
For full paper and sovereign briefings: protocol@fijishi.com