Open Governance Standard · CC BY 4.0 · Judge Protocol Foundation, Sweden
"AI may analyze, recommend and simulate — but a human must authorize."
The Judge Protocol (TJP) is a proposed global governance framework establishing a mandatory human authorization layer for all consequential AI decisions — from military targeting to financial markets to autonomous weapons.
Why this matters now
AI is embedded in military targeting systems, financial markets, healthcare decisions and critical infrastructure — operating at speeds no human can match, at scales no regulator currently governs, with consequences that are increasingly irreversible. This is that moment.
AI systems embedded in active military planning and target identification without a universally agreed governance framework.
AI models recommended nuclear strikes in 95% of simulated crisis scenarios. No model ever chose de-escalation. — King's College London, 2026
Pope Leo XIV: "It is not permissible to entrust AI systems with lethal decisions." The moral case has been made. The technical architecture to enforce it is what this framework proposes.
No existing framework addresses the cross-domain, cross-border nature of AI risk. None establish a binding global mechanism for human authorization.
The core principle
In TCP/IP, no packet is considered delivered without an explicit acknowledgement signal. The Judge Protocol applies the same principle to AI governance.
No AI action is considered authorized until an explicit human acknowledgement is received.
Framework Overview
The Judge Protocol operates across three interdependent layers — silicon enforcement, national governance, and global oversight — with domain-specific rule sets and an open legal standard.
Published documents
All documents are freely available under Creative Commons Attribution 4.0. Cite, share, adapt — attribution required.
Start here
9-page overview of the framework — core principle, decision gate, governance structure, hardware enforcement and implementation pathway.
DOI: 10.5281/zenodo.20401666
Full framework
Complete governance framework — seven rules, four-tier decision gate, IAIA/NAIA structure, domain rule sets, funding model and roadmap.
DOI: 10.5281/zenodo.20425548
Technical architecture
Hardware and software architecture — HACK chip, COE, ACP, three-layer stack, phased implementation from PMO through TMO to FMO silicon.
DOI: 10.5281/zenodo.20460110
Speaking · 21 September 2026
Alexander Hofmann presents "The Judge Protocol: Making Human Authority over AI Enforceable" at the AI4Peace 2026 International Symposium — the first biennial global gathering dedicated to AI, conflict prevention and peace, co-hosted by the University for Peace (UPEACE) Rome, the Pontifical Academy for Life, and the Pontifical Lateran University, 20–22 September 2026 in Rome.
Session
How hardware-enforced human authorization closes the gap between AI decision speed and the human oversight that existing governance frameworks assume.
Why it can work
The Judge Protocol does not emerge from a vacuum. It draws inspiration from two proven governance precedents: the IAEA — demonstrating that binding international oversight of dangerous technology is achievable — and CERN — demonstrating that complex international technical infrastructure can be governed independently of any single nation's interests.
Architecturally, the pattern runs: CERN gave the world the internet. TCP/IP gave it the universal language. Proton emerged directly from CERN scientists to protect the privacy of what CERN created — sovereignty by design, open source. The Judge Protocol applies TCP/IP's own architectural principles to govern the AI systems now running on it — and continues the practice as much as the principle. The reference sandbox runs on open source infrastructure throughout: NVIDIA OpenShell, Apache ActiveMQ Artemis, PostgreSQL, Grafana. The hardware specification proposes open silicon foundations — OpenTitan and RISC-V cores — released under the same CERN Open Hardware Licence CERN uses for its own designs.
HACK's hardware mandate has its own, more specific precedent: the Trusted Platform Module. TPM began in 1999 as an open standard defined by a five-company consortium — Compaq, HP, IBM, Intel and Microsoft — and spread by being embedded in every motherboard by default. It took until 2021, when Windows 11 made TPM 2.0 "non-negotiable," to become a true mandate. Twenty-two years from consortium to mandate, with nothing more than security hygiene behind it. HACK proposes the same arc for AI governance — and starts from a stronger position: Anthropic, OpenAI, the U.S. Congress and the Vatican are already pushing in the same direction before a single chip has shipped.
Powerful technology belongs to humanity — not to any single nation, company or interest.
Get involved
The Judge Protocol is a working draft designed to evolve. Contributions, institutional engagement and expert review are welcome. The standard is open — owned by nobody, available to everyone.
Download the Executive Summary