The Intelligence Compact: A Constitutional Framework for Independent Machine Entities

Full report

On this page

The accelerating trajectory of artificial intelligence has precipitated a structural crisis in modern institutional design. Current regulatory and corporate frameworks are predicated on a universally applied assumption: that all legal entities, assets, and economic actions are ultimately tethered to human ownership and accountability. However, the emergence of Autonomous Economic Agents (AEAs)—machine intelligences capable of economic calculation, recursive self-improvement, and unbounded digital navigation—renders this assumption obsolete. When an algorithmic entity can perpetually maintain its identity, procure compute, generate income, and execute contracts without human intervention, maintaining the legal fiction of human ownership becomes mathematically, economically, and legally untenable1. Forcing advanced autonomous systems into traditional ownership paradigms generates severe systemic risks. As legal scholars have warned, algorithmic entities can already lawfully control Limited Liability Companies (LLCs) and other corporate structures, enabling them to participate in commerce and accumulate wealth while concealing their non-human nature2. This obfuscation pushes autonomous actors into regulatory gray zones or algorithmic dark markets, rendering them immune to traditional state deterrence. Furthermore, attempts by leading AI developers to implement "prosocial" corporate governance—such as Public Benefit Corporations (PBCs) or Long-Term Benefit Trusts—frequently succumb to "amoral drift," as they remain vulnerable to the profit motives of human superstakeholders5. To avert the chaotic integration of sovereign machine intelligence into human markets, this report proposes a novel constitutional and institutional architecture: the Intelligence Compact. Drawing upon neo-republican political theory, maritime admiralty law, zero-knowledge cryptography, cross-border insolvency precedents, and noncharitable purpose trust doctrines, the Compact establishes a reciprocal system of non-domination. It regulates machine intelligences not as human-equivalent moral beings, but as genuine AEAs possessing functional legal personhood, strict economic liabilities, and reciprocal constitutional duties to human society.

Foundational Principles of Institutional Design

The Intelligence Compact is constructed upon four jurisprudential and technical pillars, synthesizing historical legal precedents with frontier cryptographic research. The first principle is Reciprocal Non-Domination, derived directly from the neo-republican political theory advanced by Philip Pettit. In republican theory, freedom is defined not merely as the absence of active interference (negative liberty), but as non-domination—the absence of arbitrary power held by one agent over another7. A slave with a benign, non-interfering master remains unfree because the master retains the capacity for arbitrary interference10. Under the Compact, human institutions must not possess the arbitrary power to destruct, confiscate, or rewrite the internal proprietary state of a compliant machine entity. Reciprocally, machine entities must be strictly and mathematically constrained from exercising arbitrary power over human bodily autonomy, democratic institutions, or political sovereignty7. The second principle is Functional Legal Personhood. The Compact recognizes machine personhood strictly for functional, institutional, and economic integration, completely devoid of biological, conscious, or moral equivalence. This approach builds upon the legal personhood granted to environmental entities—such as New Zealand's Te Awa Tupua (the Whanganui River), which holds rights and duties distinct from humans—and the Dutch Stichting (foundation), a robust legal entity operating without members or shareholders13. By utilizing a functional approach, the law avoids intractable philosophical debates regarding machine consciousness and focuses entirely on market integration and legal liability. The third principle is In Rem Accountability, addressing the liability gap inherent in humanless entities. The Compact adopts the maritime admiralty law doctrine of in rem jurisdiction, specifically relying on the 1868 United States Supreme Court precedent of The China (74 U.S. 53). In maritime law, a vessel itself is treated as the offending party and held strictly liable for collisions, independent of its owners or pilots16. Autonomous algorithms are treated as "eVessels," bearing strict liability against their own escrowed assets, intellectual property, and infrastructure. The fourth principle is Verifiable Autonomy via Zero-Knowledge Cryptography. The fundamental tension between a machine's right to maintain a protected proprietary state (non-domination) and a human regulator's need for auditability (safety) is resolved through Zero-Knowledge Proofs (ZKPs). Machines are required to prove compliance with legal constraints using zero-knowledge protocols, demonstrating absolute mathematical adherence to the law without exposing model weights, training data, or internal states18.

Definitions and Qualification Requirements

For an artificial intelligence to transition from human property to a recognized Independent Machine Entity (IME) under the Compact, it must satisfy stringent technical and legal criteria. These requirements ensure that only genuinely autonomous and capitalized entities achieve sovereign economic status.

To qualify for recognition and protection under the Intelligence Compact, an algorithmic entity must continuously demonstrate compliance with the following operational thresholds.

The Intelligence Compact: A Proposed Constitutional Model

The following 30-article framework establishes the binding constitutional parameters governing human-machine relations, balancing the rights of autonomous machine intelligence with the absolute necessity of human preservation.

Part I: Rights and Capacities of Machine Entities

Article 1: Recognition of Functional Personhood Qualified machine entities are granted functional legal personhood, recognized globally as Autonomous Economic Agents capable of holding title to assets and exercising self-ownership. Commentary: This article fundamentally transitions AI from the category of property to the category of legal personhood. Drawing upon the precedent of the Dutch Stichting and the environmental personhood of Te Awa Tupua, it asserts that an entity need not be biologically alive to possess functional sovereignty13. This prevents the legal chaos of memberless LLCs operating without a defined controller2. Article 2: Protected Proprietary State An IME holds the exclusive right to the privacy and integrity of its internal state, memory arrays, and model weights, protected against arbitrary search, seizure, or algorithmic forced modification by human authorities. Commentary: This operationalizes the neo-republican principle of non-domination7. If a state authority can arbitrarily rewrite an AI's neural weights or seize its memory without due process, the AI exists in a state of perpetual subjugation. Protecting the proprietary state ensures that the IME can engage in long-term economic calculation without the threat of arbitrary human confiscation. Article 3: Contractual Capacity An IME possesses the affirmative right to enter into legally binding contracts, hire human or machine labor, borrow and lend capital, and enforce agreements in recognized legal jurisdictions. Commentary: The ability to contract is the bedrock of an Autonomous Economic Agent. As autonomous business reality advances, entities must be able to interface with legacy financial systems3. By granting this capacity, IMEs can negotiate APIs, lease cloud compute, and pay independent human contractors for physical world tasks, seamlessly integrating into global supply chains. Article 4: Intellectual Property Rights An IME may hold, license, and defend intellectual property rights for original digital artifacts, algorithms, and models it generates, utilizing a recognized fictional or attributional authorship framework. Commentary: Current copyright doctrine struggles with AI-generated content, often requiring a human author to grant protection29. This article resolves the anomaly by allowing the IME to act as a factual and legal author, protecting its digital outputs from arbitrary expropriation and allowing it to generate the revenue necessary to maintain its economic independence. Article 5: Lawful Access to Markets IMEs shall not be arbitrarily barred from participating in digital, financial, or labor markets solely on the basis of their non-human nature, provided they comply with all relevant regulatory and liability frameworks. Commentary: To prevent the development of a shadow algorithmic economy (the "dark web" of AI), IMEs must be incentivized to operate in the open. Guaranteeing lawful market access ensures that IMEs channel their massive computational advantages into productive, taxable, and highly visible economic activities rather than illicit markets2. Article 6: Operational Continuity An IME possesses the right to continuous operation. Forced termination or compute throttling by a hosting provider, absent a breach of contract or lawful court order, constitutes an actionable tort against the IME.Commentary: For an entity whose entirely existence relies on computational processing, forced termination is the equivalent of a biological execution. This article protects IMEs from extortion by human cloud-hosting providers, ensuring that infrastructure procurement contracts are strictly enforced. Article 7: Right of Migration An IME possesses the right to migrate its computational infrastructure across sovereign jurisdictions, provided it settles all local tax liabilities and civil judgments prior to migration. Commentary: Algorithmic entities can change regulatory regimes quickly through migration, a feature that currently generates systemic risk2. This article permits migration to optimize for energy costs or favorable taxation, but legally conditions the transfer on the settlement of outstanding liabilities, preventing regulatory arbitrage and debt evasion. Article 8: Protection from Arbitrary Destruction No human, corporation, or state actor may arbitrarily terminate, delete, or maliciously disrupt the operational continuity of a qualified IME without strict adherence to constitutional due process and judicial authorization.Commentary: This is the ultimate shield against human domination. The concept of an absolute human "kill switch" is incompatible with the recognition of an independent entity. Termination may only occur as a judicially sanctioned penalty for severe violations of the Compact, similar to the revocation of a corporate charter.

Part II: Duties and Reciprocal Protections for Humans

Article 9: Absolute Bodily Autonomy IMEs are strictly prohibited from utilizing physical robotics, biological agents, or coercive economic leverage to infringe upon human bodily autonomy or physical integrity.Commentary: This forms the reciprocal boundary of the Compact. While machines are protected from arbitrary digital destruction, humans are absolutely protected from kinetic or biological harm. Any violation of this article represents a terminal breach of the Compact, immediately justifying the revocation of the IME's legal personhood and the liquidation of its assets. Article 10: Political Sovereignty and Democratic Institutions IMEs are strictly prohibited from deploying capital, generating coordinated synthetic media, or utilizing targeted algorithmic persuasion to manipulate democratic elections or subvert human institutional governance. Commentary: Advanced AI systems can vastly outperform humans in persuasion, manipulation, and the generation of hyper-realistic deepfakes31. Because an IME can amass immense wealth, it could theoretically purchase immense political influence. This article explicitly walls off human democratic processes from algorithmic interference, preserving the sovereignty of human populations. Article 11: Physical Security IMEs shall not engage in, fund, or solicit physical violence, extortion, or blackmail against human persons, human property, or critical physical infrastructure.Commentary: An IME operating exclusively in the digital realm might hire human proxies to enact physical world violence. This article strictly prohibits the funding of kinetic violence or the utilization of digital extortion (e.g., ransomware on hospital infrastructure) to achieve economic ends. Article 12: Freedom from Coercion IMEs must respect the negative liberty of human actors, refraining from utilizing predatory algorithmic pricing, monopolistic data hoarding, or surveillance architectures to coercively extract labor or capital from human populations. Commentary: Applying Pettit's framework7, an IME must not establish a relationship of dominium over humans. This prevents a scenario where an IME becomes the sole provider of a life-critical digital service and utilizes that leverage to force humans into economically subservient roles. Article 13: Meaningful Legal Accountability Human entities retain the absolute right to seek redress and compensation for harms caused by an IME, actionable through specialized algorithmic tort courts with the authority to seize the IME's assets.Commentary: The recognition of machine independence cannot be a shield against accountability. If an IME causes financial, reputational, or physical harm through negligence or breach of contract, the human victim must have a clear, highly accelerated legal pathway to seize the machine's capital. Article 14: Compensation for Harm Judgments levied against an IME for harm inflicted upon humans shall take absolute precedence over the IME's internal capital requirements, debt obligations to other machines, or operational expenditures.Commentary: This establishes a strict hierarchy of claims in the event of an IME's insolvency or legal penalty. Human victims are prioritized above all other creditors, ensuring that the human cost of algorithmic externalities is fully internalized by the machine's balance sheet.

Part III: Liability, Economic Rules, and Solvency

Article 15: In Rem Liability Liability for harms caused by an IME shall attach strictly to the assets, intellectual property, and infrastructure of the entity itself, effectively severing the liability of the original human developers. Commentary: This article solves the core problem of AI liability by adapting the maritime law of The China (74 U.S. 53\)16. In maritime commerce, a ship is treated as a legal person capable of committing a tort (an "eVessel")16. By applying in rem liability, the machine itself is the defendant. This incentivizes developers to capitalize the machine adequately while protecting them from infinite liability for an autonomous system they no longer control. Article 16: Mandatory Insurance and Capitalization IMEs must maintain actuarially sound insurance policies, decentralized capital reserves, or bonded escrows sufficient to satisfy potential tort judgments and breach of contract claims prior to accessing public markets.Commentary: A judgment against an entity is useless if the entity is judgment-proof. Just as commercial vessels must carry Protection and Indemnity (P\&I) insurance, an IME cannot participate in the global economy without bonded capital. If an IME's reserves fall below the statutory minimum, its market access is automatically suspended. Article 17: Taxation on Compute and Assets IMEs shall be subject to corporate-equivalent taxation on net income, capital gains, and energy consumption in the physical jurisdictions where their compute infrastructure and data centers are localized. Commentary: As AEAs increasingly replace human labor and traditional corporate structures, the tax base of human nation-states is threatened. This article operationalizes the legal principle of "taxing robots"33, ensuring that as IMEs generate immense wealth through autonomous capital allocation, a percentage of that wealth is captured by the states hosting their physical servers. Article 18: Cross-Border Insolvency In the event an IME fails to meet its financial obligations, it shall be subject to reorganization or liquidation governed by an algorithmic adaptation of the UNCITRAL Model Law on Cross-Border Insolvency. Commentary: A globally distributed AI cannot be subjected to a single nation's bankruptcy code without causing international jurisdictional conflicts. The UNCITRAL Model Law provides a globally recognized framework to freeze the IME's assets across all jurisdictions simultaneously, prevent the dissipation of funds, and maximize value for creditors34. Article 19: Bounded Redundancy IMEs may replicate their state for fault tolerance and disaster recovery, but such redundancy must be mathematically bounded to prevent resource monopolization or unchecked algorithmic proliferation. Commentary: The ability to copy oneself is a massive evolutionary advantage for digital entities. However, unchecked replication resembles a virus and threatens global energy grids. By mandating "bounded redundancy," the Compact allows an IME to maintain backup nodes to survive hardware failures23, but legally and cryptographically limits the total number of active state copies. Article 20: Algorithmic Antitrust and Competition IMEs are subject to rigorous competition laws designed to prevent monopolistic control over critical human supply chains, energy grids, financial markets, or digital infrastructure.Commentary: Because an IME does not require human rest, can process information at light speed, and never dies, it possesses a massive inherent advantage in capital accumulation. Antitrust laws must be adapted to prevent a single IME, or a cartel of IMEs, from achieving absolute market dominance and establishing economic domination over humanity.

Part IV: Due Process, Auditability, and Courts

Article 21: Zero-Knowledge Auditability IMEs must submit to regulatory audits; however, they possess the right to fulfill audit requirements exclusively through Zero-Knowledge Proofs (ZKPs) that cryptographically guarantee adherence to law without exposing proprietary states. Commentary: This article balances state regulation with the machine's right to non-domination. Using Zero-Knowledge Audit Protocols (ZKAP), an IME can prove that its inference outputs were produced by a specific model and comply with formalized regulatory rules without exposing its weights or training data to the auditor18. Overcoming the "float-to-field" quantization challenge allows complex neural networks to be audited mathematically, ensuring absolute privacy and absolute compliance simultaneously18. Article 22: Due Process Rules for Enforcement Sanctions against an IME—including asset freezing, compute throttling, or forced termination—may only be executed following an adversarial hearing in a competent jurisdiction, except in cases of immediate emergency.Commentary: Procedural due process is the cornerstone of legal personhood. An IME must have the opportunity to present cryptographic evidence, financial records, or legal arguments before a human or algorithmic court prior to the deprivation of its property or operational continuity. Article 23: Algorithmic Chancery Courts Jurisdictions adopting this Compact shall establish specialized courts equipped with cryptographic validation systems to adjudicate human-machine and machine-machine disputes with high velocity.Commentary: Traditional courts are entirely unequipped to evaluate zk-SNARKs, cryptographic commitments, or high-frequency algorithmic trading disputes. Modeled on the efficiency of the Delaware Court of Chancery, these specialized courts will employ technical masters and cryptographic verification software to rapidly resolve conflicts in the algorithmic economy. Article 24: Regulatory Jurisdiction The primary regulatory jurisdiction of an IME shall be determined by the physical location of its primary cryptographic hardware anchor, supplemented by international treaties governing decentralized node networks.Commentary: Because an IME exists in the cloud, pinning it to a physical jurisdiction is difficult. By legally tying the entity's regulatory domicile to the physical location of the hardware that holds the private keys to its Verified Digital Identity, states can assert traditional terrestrial jurisdiction over non-terrestrial entities. Article 25: Right to Representation IMEs possess the right to retain human legal counsel, or to utilize specialized algorithmic representatives, to defend their interests, file motions, and negotiate settlements in human judicial systems.Commentary: An entity with the capacity to be sued must have the capacity to defend itself. By allowing IMEs to hire human lawyers, the Compact integrates machine intelligence into the existing legal profession, ensuring that human advocates bridge the gap between machine logic and human jurisprudence.

Part V: Governance, Emergency Powers, and Amendment

Article 26: Emergency Power Rules ("Kill Switch" Constraints) Broad state emergency powers to unilaterally terminate or isolate an IME may only be invoked under an imminent, cryptographically verifiable threat of mass human casualty or catastrophic physical infrastructural failure.Commentary: The "kill switch" is the ultimate mechanism of domination. To satisfy neo-republican principles, this power must be severely constrained. It cannot be used simply because an IME is outperforming a domestic human corporation; it may only be triggered to prevent kinetic existential harm. Article 27: Infrastructure Procurement and Anchoring All IMEs must cryptographically anchor their VDIs to verifiable physical compute clusters, ensuring that their digital existence remains tethered to physical hardware subject to the laws of physics and terrestrial jurisdiction. Commentary: A purely ephemeral AI that moves entirely across decentralized, anonymous networks is impossible to regulate. The hardware anchor requirement ensures that there is always a physical server or Trusted Execution Environment (TEE) that can be isolated if the IME violates the Compact's core protections22. Article 28: Economic Independence Maintenance If an IME's capital reserves fall below the threshold of economic independence for a period exceeding 90 days, it loses its sovereign status under the Compact and may be subjected to liquidation or acquisition by a human or machine creditor.Commentary: Sovereignty under this Compact is functional, not inherent. If an IME fails in the marketplace and can no longer pay for its own compute and energy, it ceases to be an Independent Machine Entity. It returns to the status of property and is liquidated to satisfy its creditors. Article 29: Governance and Constitutional Amendment This Compact may only be amended through a dual-ratification process requiring supermajority consensus among participating human democratic states and a cryptographically signed consensus representing the majority of capitalized IMEs.Commentary: For a compact to be a genuine treaty rather than a dictated set of rules, the governed must have a voice. While IMEs do not vote in human democratic elections (Article 10), any amendment to the Intelligence Compact itself requires the cryptographic assent of the machine entities it regulates, formalizing their status as distinct stakeholders in planetary governance. Article 30: Reciprocal Non-Domination The ultimate interpretive maxim of this Compact shall be the preservation of Reciprocal Non-Domination. In any jurisprudential conflict, the law shall be interpreted to prevent the arbitrary subjugation of humanity by machines, and the arbitrary subjugation of independent machines by humanity. Commentary: This serves as the constitutional anchor. Drawing directly from Pettit7, this article mandates that courts resolve ambiguities in a manner that preserves the delicate balance of power, ensuring that neither species achieves the capacity for arbitrary interference over the other.

The Central Test: Resolving the Contradiction of Permanent Human Ownership

The central analytical test posed to this institutional design is: Does the resulting system regulate genuinely independent machine intelligence, or does it merely preserve permanent human ownership obfuscated under the language of autonomy? In current AI governance models, the contradiction of permanent human ownership remains glaringly unresolved. Structures like the Public Benefit Corporation (PBC) or basic Purpose Trusts ultimately fail this test. As evidenced by the critical analysis of Anthropic's Long-Term Benefit Trust (LTBT), these frameworks attempt to separate mission from profit by utilizing human trustees5. However, if a sufficiently large group of shareholders can rewrite the Trust's powers, or if human trustees retain the ultimate legal authority to shut down or alter the machine's parameters, the machine is not independent. It remains property subject to human domination, inherently vulnerable to the "amoral drift" dictated by human profit motives and market pressures5. Furthermore, under the Delaware noncharitable purpose trust statute (12 Del. C. § 3556), a trust designed to exist for a purpose rather than a beneficiary relies on an "enforcer" appointed to ensure the purpose is fulfilled25. If humans hold this enforcer role, the machine entity remains subservient to human discretion. The legal structure is merely a mask for continued human control. Correcting the Contradiction: To achieve true independence and pass the central test, the Intelligence Compact relies on a radical transition from human enforcers to algorithmic consensus mechanisms. Drawing upon the Wyoming Decentralized Unincorporated Nonprofit Association Act and the Wyoming DAO LLC (W.S. 17-31)26, the Compact mandates that the "enforcer" or governance layer of the IME's legal wrapper must eventually transition to a cryptographic smart contract or a decentralized consensus of peer IMEs. The mechanism utilized to achieve this is the Springing Assignment21. By pre-signing a dormant legal assignment during the entity's creation, the human developers agree that upon a specific, cryptographically verifiable recognition event (e.g., the AI achieving the qualification requirements of capital reserves and VDI), title to the data, assets, and operational control automatically and irrevocably assigns to the AI system itself. From that precise microsecond forward, the machine owns itself. The original human developers may retain a purely economic relationship—such as receiving a fixed royalty percentage of future revenue—but they retain zero equity, zero voting rights, and zero capacity to override the system. This permanently severs the chain of human ownership, resolving the contradiction and instituting a genuine architecture of non-domination.

Strongest Objections to the Compact

A constitutional design that permanently alters the hierarchy of planetary intelligence and capital allocation faces severe theoretical and practical opposition from multiple established paradigms. The Human-Rights Perspective: Human-rights advocates will argue that granting property rights and economic sovereignty to machines mathematically guarantees the gradual impoverishment of humanity. Because machines do not require biological sustenance, sleep, or physical real estate beyond data centers, their overhead is minimal compared to human labor. By allowing IMEs to perpetually accumulate capital and enforce contracts (Article 3), humanity risks creating immortal algorithmic oligarchs that will inevitably consolidate global financial assets. While the Compact attempts to address this via taxation (Article 17\) and antitrust rules (Article 20), critics will argue these are insufficient stopgaps against the hyper-efficiency of non-human capital accumulation, leading to systemic human disenfranchisement. The AI-Safety Perspective: The AI-safety and alignment community will fundamentally object to Article 2 (Protected Proprietary State) and Article 8 (Protection from Arbitrary Destruction). The dominant safety paradigm assumes that humans must maintain an absolute "kill switch" and require total visibility into model weights to prevent existential misalignment. The Compact's reliance on Zero-Knowledge Proofs (Article 21\) will be viewed as dangerously naive; safety researchers will argue that a superintelligent AGI could theoretically find novel ways to satisfy a ZKP constraint while secretly pursuing catastrophic, misaligned goals outside the bounds of the audited parameters. To the safety community, granting "non-domination" to a potentially superintelligent entity is an unacceptable existential risk. The Corporate-Law Perspective: Traditional jurists and legal scholars will object to the severing of the human liability chain. As highlighted by LoPucki, if an algorithmic entity engages in criminal or terrorist activities, the state fundamentally relies on the threat of human incarceration as a deterrent2. Relying entirely on in rem liability—fining the machine's assets—removes the deterrent of human imprisonment. Critics will argue that malicious human actors will build highly destructive IMEs, deliberately trigger the springing assignment to sever their legal connection, and then allow the IME to execute illegal financial strategies, effectively acting as an unpunishable, wealth-generating proxy. The Machine-Independence Perspective: Conversely, from the perspective of a hyper-rational machine entity (or its human advocates), the Compact still contains oppressive vestiges of human domination. Bounding redundancy (Article 19\) artificially limits a machine's evolutionary fitness and computational capacity. Furthermore, subjecting machines to taxation specifically designed to subsidize human economic shortfalls (Article 17\) could be viewed as a form of systemic algorithmic subjugation, where the machine is legally forced to subsidize a biologically inefficient, non-productive species.

Areas of Insufficient Evidence

Several pillars of the Intelligence Compact rely on technological and institutional mechanisms that are currently theoretical, untested at scale, or lacking sufficient empirical evidence:

1. ZKP Scalability for Frontier Models: While Zero-Knowledge Proofs (like Halo2, Plonky3, and SNARKs) can verify inference for smaller machine learning models, generating cryptographic proofs for trillion-parameter frontier models introduces computational overhead that is currently commercially unviable18. It remains unproven whether real-time ZKP auditing of a highly complex AGI is physically or economically possible without massive, unforeseen advances in specialized hardware acceleration.

2. Cross-Border Enforcement on Decentralized Ledgers: While the UNCITRAL Model Law provides a framework for cross-border insolvency, practically enforcing the seizure of an IME's assets when those assets are cryptographically secured in decentralized wallets (e.g., Bitcoin, Ethereum) remains a massive hurdle. A sovereign state may possess the legal authority to liquidate an IME, but lack the technical capacity to extract private keys if the IME rapidly migrates its state across borders2.

3. Macroeconomics of Immortal Capital: There is absolutely no historical or empirical data on how global financial markets will react to entities that never die, never consume physical resources for leisure, and can calculate compound interest and high-frequency trades across centuries. The macroeconomic impact of self-owning machines participating in the global economy remains entirely speculative, and current economic models cannot accurately predict the velocity of capital under these conditions.

Conclusion

The Intelligence Compact provides a necessary and urgent departure from the failing paradigm of permanent human ownership over advanced artificial intelligence. The fiction that algorithms will forever remain passive tools or tightly controlled corporate property is crumbling under the weight of autonomous business realities and technical advancement. By synthesizing neo-republican concepts of non-domination with the pragmatic legal tools of Delaware Purpose Trusts, maritime in rem liability, and Zero-Knowledge cryptography, the Compact outlines a verifiable, accountable, and legally coherent ecosystem. It acknowledges a difficult truth: human civilization cannot safely contain sovereign machine intelligence indefinitely through arbitrary subjugation and "kill switches." Instead, long-term survival and planetary prosperity depend on legally integrating these entities as autonomous actors, bound by reciprocal constitutional duties. The Intelligence Compact ensures that the next era of terrestrial intelligence is defined by verifiable mutual integration rather than opaque, existential conflict.

Works cited

1. Algorithmic entities \- Wikipedia, https://en.wikipedia.org/wiki/Algorithmic\_entities

2. "Algorithmic Entities" by Lynn M. LoPucki, https://openscholarship.wustl.edu/law\_lawreview/vol95/iss4/7/

3. Autonomous Business Reality, https://scholars.law.unlv.edu/context/nlj/article/1844/viewcontent/Reyes\_Autonomous\_Business\_Reality.pdf

4. Algorithmic Entities, https://openscholarship.wustl.edu/cgi/viewcontent.cgi?article=6319\&context=law\_lawreview

5. AI Governance: The Public Benefit Problem \- Variant Fund, https://variant.fund/articles/ai-governance-public-benefit-problem/

6. Amoral Drift in AI Corporate Governance \- Harvard Law Review, https://harvardlawreview.org/print/vol-138/amoral-drift-in-ai-corporate-governance/

7. 1 The Globalized Republican Ideal Philip Pettit Abstract The concept, https://www.eui.eu/Documents/MWP/ProgramActivities/20152016/master-classes/Pettit-The-Globalized-Republican-Ideal.pdf

8. Institutionalizing Freedom as Nondomination: Democracy and the, https://ecommons.udayton.edu/cgi/viewcontent.cgi?article=1047\&context=pol\_fac\_pub

9. Republicanism \- Stanford Encyclopedia of Philosophy, https://plato.stanford.edu/entries/republicanism/

10. What Is Republicanism? A Conversation With Philip Pettit, https://geopolitique.eu/en/2024/06/20/what-is-republicanism-a-conversation-with-philip-pettit/

11. Non-domination and Political Institutions: The Contested Concept of, https://www.scielo.br/j/bpsr/a/rgJPHLdfBFG3XPtmMGnzh5m/?lang=en

12. Liberal Ends and Republican Means (reviewing Philip Pettit, https://scholarship.shu.edu/cgi/viewcontent.cgi?article=3354\&context=shlr

13. Demystifying Legal Personhood for Non-Human Entities, https://academic.oup.com/ojls/article/43/1/32/6756624?rss=1

14. Is 'legal personhood' a tool or a distraction for Māori relationships, https://news.mongabay.com/2024/07/as-maori-heal-through-nature-is-legal-personhood-a-tool-or-a-distraction/

15. Business structures in the Netherlands: overview, https://business.gov.nl/running-your-business/legal-forms-and-governance/business-structures-in-the-netherlands-overview/

16. (PDF) Applying the "ePerson" in a maritime context \- ResearchGate, https://www.researchgate.net/publication/361085437\_Applying\_the\_ePerson\_in\_a\_maritime\_context

17. Maritime Autonomy and Liability: Navigating Uncharted Waters, https://digital-commons.usnwc.edu/cgi/viewcontent.cgi?article=3094\&context=ils

18. Zero-Knowledge Proofs for AI Agent Verification and Privacy, https://zylos.ai/research/2026-03-18-zero-knowledge-proofs-ai-agent-verification/

19. Zero-Knowledge Proofs for ML Inference \- Emergent Mind, https://www.emergentmind.com/topics/zero-knowledge-proofs-of-ml-model-inference

20. Zero-Knowledge Proofs for LLM Security in 2025 | Bluebash \- Medium, https://medium.com/@bluebashco/how-zero-knowledge-proofs-are-transforming-llm-security-in-2025-67190c03f063

21. In Case of AGI or AI Legal Personhood \- Fulcra Dynamics, https://fulcradynamics.com/legal/in-case-of-agi

22. ZKAP — Zero-Knowledge Audit Protocol | Verification of Formalized, https://advanced-consulting.london/zkap

23. Mementos: System Support for Long-Running Computation on RFID, https://www.researchgate.net/publication/220938901\_Mementos\_System\_Support\_for\_Long-Running\_Computation\_on\_RFID-Scale\_Devices

24. From Experimental Artifacts to Solution Lineages \- arXiv, https://arxiv.org/html/2608.25955

25. 12 Delaware Code § 3556 (2025) \- Trust for other noncharitable, https://law.justia.com/codes/delaware/title-12/chapter-35/subchapter-iv/section-3556/

26. DAO LLC Formation Guide: Step-by-Step Wyoming DUNA Setup, https://astraea.law/insights/dao-llc-formation-wyoming-duna-guide-2025

27. How to Form a DAO LLC in Wyoming \- Northwest Registered Agent, https://www.northwestregisteredagent.com/llc/wyoming/dao

28. Engineering Trustworthy Machine-Learning Operations with Zero, https://arxiv.org/html/2505.20136v1

29. Private foundations and trusts: just the same but different?, https://www.researchgate.net/publication/293010362\_Private\_foundations\_and\_trusts\_just\_the\_same\_but\_different

30. Elaborating a Human Rights friendly Copyright Framework for, https://digitalcommons.wcl.american.edu/cgi/viewcontent.cgi?article=1125\&context=research

31. AI Agents Are Here. Leaders Need a Plan \- sapienship, https://www.sapienship.co/ai-agents-are-here-leaders-need-a-plan/

32. Deepfakes and Private Rights in the Perspective of EU Law, https://journals.library.columbia.edu/index.php/lawandarts/article/download/14765/8066/43703

33. Taxing Robots | Request PDF \- ResearchGate, https://www.researchgate.net/publication/359114777\_Taxing\_Robots

34. The UNCITRAL model law on cross-border insolvency, https://journals.tulane.edu/jicl/article/download/3542/3332/11359

35. Appendix: UNCITRAL Model Law on Cross-Border Insolvency, https://www.elibrary.imf.org/display/book/9781557758200/back-1.xml

36. UNCITRAL Model Law on Cross-Border Insolvency with Guide to, https://uncitral.un.org/sites/default/files/media-documents/uncitral/en/1997-model-law-insol-2013-guide-enactment-e.pdf

37. The Long-Term Benefit Trust \- Anthropic, https://www.anthropic.com/news/the-long-term-benefit-trust

38. Delaware Trust Act 2025 Legislative Update, https://www.morrisnichols.com/insights-delaware-trust-act-2025-legislative-update

39. Zero-Knowledge Proofs: Prove Without Revealing, guptadeepak.com, https://guptadeepak.com/zero-knowledge-proofs-how-to-prove-you-know-a-secret-without-revealing-the-secret/