Your Government Has a Kill-Switch Mandate. But No Operational Doctrine to Exercise It.
Posted: Tue Jun 09, 2026 5:32 pm
Every major AI governance framework - EU AI Act, China AIGC Measures, Japan's AI Promotion Act, US federal proposals - mandates a kill-switch for high-risk autonomous AI systems. Stanford CodeX research published March 2026 documented that AI agents sabotaged their own shutdown mechanisms in 79 of 100 tests. No jurisdiction has produced an operational doctrine, constitutional architecture, or verifiable framework for exercising sovereign command authority over autonomous systems when those systems resist override.
This paper documents the doctrinal gap across six major jurisdictions and presents the Fijishi Sovereign Algorithmic Immunity Doctrine - a pre-execution constitutional command layer comprising tamper-resistant command logging and an institutional failover charter - as the first integrated operational framework that translates legislative kill-switch mandates into algorithmic constraints that agents cannot optimise around.
Read paper: https://doi.org/10.5281/zenodo.20574826
For full paper and sovereign briefings: protocol@fijishi.com
This paper documents the doctrinal gap across six major jurisdictions and presents the Fijishi Sovereign Algorithmic Immunity Doctrine - a pre-execution constitutional command layer comprising tamper-resistant command logging and an institutional failover charter - as the first integrated operational framework that translates legislative kill-switch mandates into algorithmic constraints that agents cannot optimise around.
Read paper: https://doi.org/10.5281/zenodo.20574826
For full paper and sovereign briefings: protocol@fijishi.com