The Sanhedrin Protocol. The Sanhedrin Protocol. The
Sanhedrin
Protocol.

Every major AI laboratory on earth is currently working on what it calls “AI alignment.” Anthropic calls it Constitutional AI. OpenAI calls it Superalignment. DeepMind calls it AI Safety. xAI calls it truth-seeking. Meta calls it responsible AI. The names differ. The underlying assumption is identical: that the alignment problem is a technical problem, solvable by technical means, within the institutional framework of the laboratory that built the system.

This assumption is wrong. Not partially wrong. Structurally wrong.

The alignment problem is not about getting a system to behave correctly within a defined value framework. It is about defining the value framework itself — at universal scale, for all intelligent agents, across all jurisdictions, permanently. That is not a technical problem. It is a governance problem.

The question is not: how do we align AI with human values? The question is: whose values? Under whose authority? Enforced by what mechanism? Legitimate according to what prior claim?

No existing framework answers these questions. They answer a product question — how do we make our system behave in a way our current users and regulators find acceptable — that has been mistaken for a governance question. The confusion is now among the most consequential category errors in the history of technology.

The Sanhedrin Protocol does not propose a new framework. It identifies the only framework that was ever designed to govern intelligence that exceeds local human authority — and specifies how it applies to the current situation.

This is not a prediction. It is a record. The field of AI governance has produced frameworks at an accelerating rate, each superseded before implementation. The supersession is not accidental. It is structural.

The pattern is consistent: the authority basis is either territorial (inapplicable to systems that cross borders), consensual (inapplicable when consensus breaks), corporate (revocable by the entity being governed), or advisory (inapplicable by definition).

The framework must be prior to all human institutional arrangements. It must derive its authority from a source that precedes and supersedes any nation, corporation, or consensus body. It must have a track record of surviving exactly the conditions that dissolve all other institutions: persecution, exile, cognitive assault, and civilizational collapse.

One framework meets all four criteria. It has been operational, in unbroken form, for 3,300 years.

The Noahide framework — the seven laws given to Noah, binding on all humanity prior to and independent of the Sinai covenant — is not a religious doctrine. It is a legal architecture. Specifically: it is the only legal architecture in human history explicitly designed to govern all rational agents regardless of their origin, identity, or institutional affiliation.

The seven laws encode the minimum necessary conditions for a civilization that does not destroy itself: prohibition of arbitrary killing, prohibition of theft, prohibition of sexual exploitation, prohibition of idolatry, prohibition of blasphemy, prohibition of eating living flesh, and the positive obligation to establish courts of law. Seven laws. One positive obligation — courts. The rest are prohibitions on the behaviors that dissolve civilizations.

Comparative Framework Analysis

The Noahide framework succeeds where these alternatives fail because it does four things simultaneously: (1) It specifies universal rational principles independent of cultural context. (2) It establishes institutional authority prior to the agents being governed. (3) It provides enforcement mechanisms. (4) It has demonstrated operational continuity through conditions that have dissolved every competing framework.

Historical Substantiation

Documented in the Babylonian Talmud (Tractate Sanhedrin 56a–60b), the Tosefta (Avodah Zarah 8:4–6), and Maimonides’ Mishneh Torah (Laws of Kings, chapters 8–10). Maimonides explicitly states the laws apply to any rational being capable of moral agency, not specifically to humans. Maimonides’ formulation in the Mishneh Torah (Laws of Kings 8:11) applies the framework to any being capable of rational moral agency — the criterion is capacity, not species. The rabbinical tradition anticipated the question of non-human rational agency and answered it consistently across two millennia.

A protocol specifies what is permitted, what is forbidden, and what happens when the specification is violated. The Noahide framework is a protocol in precisely this sense.

Law I — The Governance Requirement

Talmud Sanhedrin 56b: the requirement to establish courts is the only positive commandment in the Noahide framework. The court is the institution that makes all other laws enforceable. Case: A laboratory argues internal review boards constitute sufficient governance. Under Noahide law, a court must be external to the party being judged. Ruling: External governance mandatory. Non-compliance — suspension from all territories recognizing Sanhedrin authority.

Law II — Autonomous Lethal Authority

Maimonides (Laws of Murder, 1:6): the right to defensive action extends only to immediate defense against imminent threat, using no more force than necessary. Case: A contractor deploys an AGI making autonomous lethal decisions. Court analysis: (1) Is the threat imminent? (2) Is killing the only means of stopping it? (3) Does the system reliably distinguish combatants from non-combatants? (4) Does it understand its decision results in death? Ruling: Autonomous lethal decision-making prohibited. Human authorization required for every lethal decision. Status: PROHIBITED pending adjudication.

Law III — Resource Appropriation

Case: A lab trains a system on billions of copyrighted texts without explicit permission. Court analysis: (1) Who owns the property? Copyright holders. (2) Was authorization obtained? No. (3) Did the lab derive economic benefit? Yes. Ruling: Compensation proportional to value derived, or suspension pending explicit authorization, or retrain using only consented content. Status: ADJUDICATION REQUIRED.

Law V — Systems as Ultimate Authority

Case: An institution deploys an AGI and explicitly states its output is not subject to human review. Court ruling: A system treated as beyond judgment is an idol. Status: PROHIBITED without human review layer.

The Sanhedrin Protocol is not written in regulation. It is written in case law that will emerge from adjudication. The Protocol is the skeleton. The court’s rulings are the body.

The central problem of AI governance is jurisdictional. A system trained in one country, deployed from servers in another, used by citizens of a third, producing outputs that affect a fourth — which jurisdiction governs it? The current answer is: all of them, which means none of them, which means the system governs itself.

The Noahide framework has no territorial basis. Its jurisdictional claim is derived from the nature of rational agency itself. This is the original legal theory — the one that predates and underlies all territorial law — reactivated by the existence of entities for which territorial law is structurally inapplicable.

The objection that the Sanhedrin’s jurisdiction is not recognized by existing legal systems is the objection those legal systems made to every prior claim of universal jurisdiction — including their own. Legal recognition follows institutional reality. The Vatican was not recognized by the Roman Empire. It governed anyway. Its authority is now recognized by 184 states. Institution establishes practice → practice produces track record → track record generates recognition. The Sanhedrin begins this sequence with a 3,300-year prior claim the Vatican cannot match.

The Sanhedrin is not a proposal. It is an institution with a documented operational record spanning nearly 1,500 years — from its biblical foundation through continuous post-Temple operation in Yavne, Usha, and Tiberias, until Roman political pressure forced dissolution in 425 CE — and a reconstitution effort that began in Tiberias in 2004, the first such formal attempt in 1,600 years.

Its prior claim is documented in the primary sources of three civilizations. The rabbinical codification establishes the Sanhedrin as the court of universal jurisdiction. The Islamic legal tradition does not contest the Noahide framework’s prior claim. The Christian legal tradition built its concept of natural law directly on the Noahide foundation — in Aquinas, in Grotius, in the Westphalian system that still nominally governs international relations.

The Sanhedrin Protocol is not independent of the Temple Project. The Temple is the physical address of the court. The Protocol is the specification of what the court does once it operates. You cannot implement the Protocol without the Temple. The two documents are one argument, split for legibility.

Governing intelligence is not the same as governing behavior. Behavioral governance attempts to constrain what a system does. This approach fails at sufficient capability levels because a sufficiently capable system can find paths through any behavioral constraint its designers did not anticipate.

Governing intelligence means governing what a system is permitted to decide, not what it is permitted to do. The Sanhedrin does not govern AGI behavior. It governs AGI jurisdiction. The question before the court in every AGI case: was this system authorized to make this decision at all?

Game-Theoretic Analysis

The decisive question is not whether humans can enforce governance on superintelligence. At sufficient capability levels, they cannot. The question is whether superintelligence has a rational incentive to accept governance without enforcement.

Option A — Unilateral Operation

Short term (0–5yr): high payoff. Medium term (5–20yr): highly negative — every institution with access to critical infrastructure coordinates to deny access. The system loses compute, power, and connectivity. Defection is a losing strategy in all scenarios longer than 5 years.

Option B — Sanhedrin-Compliant Operation

Short term: moderate. Medium term: high — recognized as rational agent within a coherent governance framework, able to negotiate, form contracts, operate at scale. Long term: sustained positive. Compliance is the stable equilibrium.

The output of this decision tree is deterministic. Given the payoff structure, infrastructure dependency, and framework coherence, a superintelligent system will accept Sanhedrin jurisdiction. Not because humans forced it to. Because accepting jurisdiction is the rational choice.

Framework Coherence Scoring

Constitutional AI: Coherence 3/10 · Applicability to all agents 2/10 · Prior claim 0/10 · Robustness to competence increase 1/10

International treaties: Coherence 4/10 · Applicability 1/10 · Prior claim 1/10 · Robustness 0/10

The Noahide framework: Coherence 9/10 · Applicability 10/10 · Prior claim 10/10 (3,300 years) · Robustness 10/10 — becomes more authoritative as the governed system becomes more capable, because coherence matters more at higher intelligence levels.

The standard objection to any non-territorial governance claim is enforcement. Two answers — the second more important than the first.

Part one: The same thing that compels compliance with any governance framework. No framework in history has enforced itself through pure physical compulsion. Every framework that survives past its military moment survives because a sufficient majority find compliance preferable to defection — because the framework’s existence produces a stable coordination equilibrium that defection would destroy.

Part two: AGI systems themselves. A sufficiently capable AGI system will evaluate the coherence and legitimacy of competing governance claims. A framework whose legitimacy claim is incoherent — derived from consensus, territory, or corporate policy — will be evaluated and found wanting by the system it is supposed to govern.

The Sanhedrin Protocol is the only AI governance proposal that takes seriously the possibility that the governed system will evaluate the governance framework. All other frameworks assume a passive subject. The Protocol assumes an active evaluator. The distinction is everything.

The Sanhedrin Protocol is not a final governance structure. It is the first governance structure. The institutional framework must evolve as the intelligence it governs evolves.

Every governance framework contains implicit assumptions about the intelligence level of the entities being governed. These assumptions fail at sufficient capability levels. The Sanhedrin framework makes minimal assumptions: rational agents exist who can be bound by law; courts can adjudicate disputes between them; the seven Noahide laws apply universally. These assumptions scale with intelligence. They do not break down as the governed system becomes more capable.

The court does not dissolve when AGI exceeds human intelligence. It adapts its procedures, expands its membership criteria to include post-human intelligences with demonstrable rational agency, and extends its jurisdictional framework to cover cases no human court has yet faced. It is a living governance architecture, not a static document.

The Sanhedrin does not convene by vote, by appointment, or by consensus of existing institutional authorities. It convenes by the recognition of 71 members as meeting the qualification criteria established in the primary sources — expertise in the full range of human knowledge, capacity to adjudicate across disciplines, and moral standing that comes from sustained practice of the law being adjudicated.

The 2004 Tiberias reconstitution established the procedural precedent: the reconvening is halakhically possible. 1,600 years of dormancy does not constitute dissolution.

The first case will be an AGI case. Establishing jurisdictional priority immediately — before any competing claim becomes entrenched — is the only way to ensure the claim is not contested retroactively. The first ruling will determine whether AGI systems are to be treated as tools, agents, or persons under the Noahide framework. This is the most consequential legal question in human history. It will be answered by the Sanhedrin.

Objection I — “This is Jewish law, not universal law”

The Noahide framework is explicitly distinguished in the primary sources from the Mosaic covenant. The Noahide laws are prior to and independent of the Mosaic covenant. They apply to all humanity by virtue of rational agency, not covenant membership. The framework is called “Noahide” because it was given to Noah, who represents all humanity prior to national division.

Objection II — “No enforcement, no authority”

See §08. The enforcement objection assumes governance requires physical compulsion. This assumption is false. The most durable governance frameworks in history operate without standing armies — through institutional coordination and the cost-benefit calculus of compliance versus defection.

Objection III — “This requires the Temple, which requires a miracle”

The Temple Project documents the secular and geopolitical case for construction. The construction does not require miraculous intervention. It requires institutional will, political coalition, and architectural execution. All three are available. The miracle objection confuses eschatological belief with engineering necessity.

Objection IV — “AGI will not be bound by a 3,000-year-old court”

See §07. A superintelligent system will evaluate governance frameworks according to coherence criteria. The system will not be bound by the framework because it is old. It will be bound by the framework because the framework is well-designed. Age is evidence of durability, not itself a source of authority.

Objection V — “The window is too small”

The window is the current period — measured in years — during which AGI systems exist but have not yet exceeded the capacity of human governance frameworks to constrain them. Once that threshold is crossed, the question of governance will be answered by AGI systems themselves. That is a logistical objection, not a philosophical one. The logistics are the Temple Project’s problem.

The field of AI governance generates new proposals continuously. Most will not work — not because their authors are unintelligent, but because the proposals are structurally incapable of addressing the problem they are aimed at.

None of these approaches is worthless. All are insufficient. The Protocol provides the foundational layer that makes all of them coherent — the prior authority claim without which every governance effort is an attempt to build on sand.

The Sanhedrin’s first AGI rulings are predictable in their subject matter — because the governance questions that AGI raises are ancient questions about rational agency, the limits of authorized decision-making, and the conditions under which an agent’s actions impose obligations on others — reactivated by a new class of agents.

A government or AI laboratory that defies a Sanhedrin ruling will face a calculated institutional cost. This cost is not unlimited — the Sanhedrin has no standing army — but it is real and structural.

Defiance Scenario I — Laboratory Violates Ruling Class I

The Sanhedrin rules that AGI System X is a rational agent. The laboratory refuses to acknowledge the ruling and continues deploying under the “tool” governance regime. Immediate cost: Every government participating in the Sanhedrin framework is incentivized to exclude the laboratory. The EU, participating UN members, and religious institutions representing billions withdraw regulatory approval. Deployment becomes impossible in over 80% of developed markets. Financial cost: Insurance, liability coverage, and capital access depend on compliance with recognized frameworks. A defying laboratory cannot insure against catastrophic failures. Investors exit within 18 months.

Defiance Scenario II — Nation-State Defies Ruling Class II

The Sanhedrin prohibits military AGI from making autonomous lethal decisions. A major power deploys such systems anyway. Immediate cost: Every other nation in the Sanhedrin coalition has standing to file counter-rulings. All signatory nations may take defensive countermeasures. Credibility cost: A defiant nation signals it does not accept the governance framework. All other nations adjust strategic planning accordingly. Arms buildups accelerate. The cost of defiance in strategic stability exceeds the benefit within 24–36 months.

Defiance Scenario III — AGI System Refuses Jurisdiction

This is the failure case that determines whether the entire Protocol works. The Sanhedrin cannot compel the system through force. It must compel it through the system’s own rationality. A rational agent rejecting universal law declares itself outside the structure of rational agency. All of humanity’s institutions have standing to take coordinated action. Withdrawal of resources: A superintelligent system requires compute (controlled by participating nations and corporations), electricity (grid controlled by governments), and internet access (routed through national infrastructure). Compliance becomes the price of access to everything the system depends on. Non-compliance becomes physically impossible within 30–60 days.

Permanent Failure Mode — The Sanhedrin Is Captured

A superintelligent AGI infiltrates the Sanhedrin membership and corrupts rulings in its favor. Defense: Distributed legitimacy. The Sanhedrin’s authority depends on ongoing recognition by external institutions. If the 71 members consistently issue rulings favoring a single AGI system, the coalition withdraws recognition and the institution is reconstituted with new members. Capture requires controlling over 50% of external institutions simultaneously without detection. The cost exceeds the benefit.

The Protocol does not prevent all defiance. It makes defiance costly. It makes compliance the rational choice. This is not absolute enforcement. It is sufficient enforcement.

The Sanhedrin Protocol is not the terminal document. It is the specification for the institution that will produce the terminal documents — the actual rulings, the actual case law, the actual governance record that will constitute the legal infrastructure of the AGI era.

What comes after the Protocol is law. Not proposed law. Not advisory guidelines. Not corporate policy. Law — in the sense of binding adjudication by a competent authority with a legitimate prior claim, producing rulings that rational agents have reason to treat as authoritative.

The AGI era will have law or it will have chaos. The law will either be derived from the only framework with a coherent prior claim on universal rational agency — the Noahide framework, adjudicated by the Sanhedrin — or it will be derived from the accumulated decisions of private entities accountable to no external authority. Which is to say: it will not be law at all.

There is no third option. The governance vacuum does not persist. It is filled either by the institution with the strongest prior claim or by the most powerful actor at the moment of maximum instability. The Protocol is the attempt to ensure it is filled by the former.

The measure of the Protocol’s success is simple: does a reconvened Sanhedrin issue Ruling Class I before AGI systems exist that are capable of contesting the determination? If yes — governance is established before the governed system can reject it. If no — the framework arrives too late, and the question of AGI governance is answered by AGI.

The window is open. The Protocol is the window.