I have been reading, with no small measure of admiration, a book entitled A Peopleβs History of the Supreme Court by Peter Irons, and I would encourage every reader of this essay to seek it out, for it is among the most comprehensive and illuminating accounts of how the highest court in the land has shaped, and been shaped by, the evolving aspirations of the American republic. I first encountered Professor Irons through his teaching courses on the history of the Supreme Court, and what began as a matter of intellectual curiosity has since become a deeply personal journey through the foundations of constitutional law, the great cases that tested those foundations, and the remarkable men and women whose arguments and dissents gave living substance to what might otherwise have remained mere parchment and ink. It is this journey, this sustained encounter with the origins and architecture of American governance, that has compelled me to draw what I believe is a necessary and urgent parallel between the constitutional struggles of the late eighteenth century and the governance challenges that now confront us in the age of artificial intelligence.
The Lesson of the Confederation
To appreciate the magnitude of what the framers accomplished in Philadelphia, one must first reckon with the depth of the failure that preceded it. The Articles of Confederation, ratified in 1781, represented the young nationβs first attempt at collective self-governance, and by nearly every consequential measure, that attempt proved inadequate to the demands of a growing and contentious union. Under the Articles, each state retained near-total sovereignty, printing its own currency, negotiating its own commercial terms, and operating under no binding obligation to honor the agreements of its neighbors. There was no executive authority to coordinate national affairs, no federal judiciary to arbitrate disputes between the states, and no mechanism by which the confederation could levy taxes or enforce the very trade agreements upon which commercial life depended. What the Articles produced was not the ordered liberty that the revolution had promised but rather a fragmentation of purpose so thoroughgoing that, within half a decade, the republic itself stood in genuine peril of dissolution.
It was this recognition, arrived at through bitter experience and no small amount of political courage, that brought the delegates to the Constitutional Convention of 1787. They gathered not to amend the Articles but to replace them with something altogether more durable, more principled, and more capable of binding a diverse confederation of interests into a coherent and functioning whole. The Constitution that emerged from those deliberations, soon fortified by the Bill of Rights, established the non-negotiable principles upon which the republic would stand: the protection of property, the guarantee of due process, the safeguarding of individual liberty against the overreach of concentrated power. These were not aspirational sentiments offered for voluntary adoption. They were structural commitments, the load-bearing walls of a republic, and upon those walls an entire economy, legal system, and civil society would be constructed over the course of the next two centuries, precisely because citizens, merchants, and institutions could rely upon the predictability and enforceability of a shared constitutional order.
A Confederation of Algorithms
I submit to the reader that we find ourselves, at this very moment, in a period of technological history that bears a striking and unsettling resemblance to those pre-constitutional years. The landscape of artificial intelligence today is a confederation in the truest and most cautionary sense of the word. Every major technology company is building its own AI agents, its own large language models, its own autonomous systems, and deploying them into domains of extraordinary consequence, from healthcare and finance to legal services, logistics, hiring, and national defense. These digital entities are not passive instruments awaiting human instruction. They make decisions. They transact. They interact with human beings and, increasingly, with one another in ways that carry material weight in the lives of real people.
And yet, for all this proliferation of autonomous capability, there exists no common standard that binds them, no shared constitutional order that governs how they must behave. Each company operates under its own internal guidelines, its own safety protocols, its own self-authored definition of what βresponsible AIβ means. Some organizations invest heavily in alignment research, red-teaming, and transparency. Others rush products to market with little more than a terms-of-service document and a carefully worded disclaimer. The result is a confederation of technologies, powerful and proliferating and dangerously fragmented, in which the rules that govern one system bear no necessary relationship to the rules that govern another, even when those systems interact in consequential ways.
Consider the practical implications of this fragmentation. When an AI agent operating within one companyβs ecosystem engages with an AI agent built by another, when they exchange data, execute transactions, or collaborate on decisions that affect the welfare of human beings, whose rules apply? What recourse does a person have when an autonomous system produces a consequential error and no single entity can be identified as clearly and unambiguously accountable? Under the Articles of Confederation, a merchant in Virginia had no reliable means of enforcing a contract with a buyer in Massachusetts, because the currencies did not align, the courts did not cooperate, and the incentives of each sovereign state diverged in ways that made cross-border commerce an act of faith rather than an act of law. We resolved that problem by establishing a constitutional framework. The question now before us, and one whose urgency grows with every new deployment of autonomous AI, is whether we possess the clarity and the collective will to do the same for the governance of artificial intelligence.
The Pillars of a New Constitutional Order
To speak of a constitution for AI is not to propose that we transcribe the clauses of the American founding document and affix them to machine learning models. It is, rather, to insist that what this moment demands is not another voluntary framework, not another set of industry principles published in a white paper that no institution is obligated to follow, but a foundational architecture of governance built upon a small number of non-negotiable commitments that every AI system, regardless of its maker, its market, or its national origin, must honor.
One might envision this architecture as resting upon several essential pillars, each of which finds its antecedent in the protections that the founders encoded for human citizens.
The first is transparency as due process. Just as the Fifth Amendment guarantees that no person shall be deprived of life, liberty, or property without due process of law, so too must AI systems that render consequential decisions about credit, employment, medical treatment, or legal risk be required to furnish explanations that allow a human being to understand why a particular decision was reached and to mount a meaningful challenge when that decision is unjust.
The second is the treatment of data rights as property rights. The Fourth Amendment protects the citizen against unreasonable search and seizure, and in the digital era, where personal data constitutes a form of personal property more intimate than any physical possession, a constitutional approach to AI governance would establish clear and enforceable boundaries around what data a system may collect, retain, and act upon, while granting individuals genuine rights of consent, access, and deletion.
The third is accountability as the rule of law. The Constitution did not merely proclaim rights in the abstract. It established institutions, courts, legislatures, and an executive, whose purpose was to give those rights the force of law. AI governance without enforcement is ornamentation. There must exist credible mechanisms to audit autonomous systems, to adjudicate disputes arising from their operation, and to impose meaningful consequences when those systems cause demonstrable harm. The current patchwork of self-regulation amounts to asking each state to police itself, and the history of the confederation tells us plainly how such arrangements conclude.
The fourth is interoperability as the commerce clause. Among the most consequential provisions of the Constitution was the commerce clause, which prevented individual states from erecting trade barriers against one another and thereby preserved the conditions under which a national economy could flourish. In the AI ecosystem, interoperability standards serve precisely the same structural function. If agents built by different companies cannot communicate transparently, if there exists no shared protocol for trust verification and data exchange, the digital economy will fracture into walled fiefdoms, each sovereign unto itself and each incapable of participating in the broader commerce of intelligence that the age demands.
The Necessity of Debate
It would be naive to suppose that any of these will be accomplished without profound and sustained disagreement. The Constitutional Convention of 1787 was not a genteel seminar. The delegates argued with ferocity over representation, over the balance between federal authority and state sovereignty, over the very nature of the republic they were attempting to construct. The Bill of Rights itself was not part of the original document but was added through political struggle, because the Anti-Federalists understood, with a prescience that history has repeatedly vindicated, that a constitution without explicit protections for individual liberty is a constitution that can become an instrument of the very oppression it was designed to prevent.
The debate over AI governance will prove no less contentious. Technology companies will resist binding standards that constrain their competitive advantage. Nations will have differing views on jurisdiction, the appropriate balance between innovation and caution, and whether AI systems should be considered tools, agents, or something entirely new under the law. Civil society will advocate for protections that industry deems burdensome, and regulators will labor to keep pace with systems that evolve more rapidly than the legislative process was ever designed to accommodate.
This friction is not an impediment to good governance. It is the very mechanism through which good governance is forged. The American constitutional process succeeded precisely because it was difficult, because competing interests were compelled to negotiate, to compromise, to articulate and defend their values under the pressure of genuine disagreement. What emerged from that crucible was imperfect, deeply and painfully imperfect in ways that would require centuries and a civil war to begin correcting, yet it proved durable enough to absorb shocks, to adapt to circumstances the framers could never have foreseen, and to hold the center of a republic while its edges continued to evolve.
From Principle to Particulars
I anticipate that readers of a practical disposition will ask, quite reasonably, what this constitution for AI governance looks like in its specifics, and I want to address that question with the candor it deserves. The honest answer is that the specifics do not yet exist in finished form, nor should we expect them to, for the same reason that the delegates who gathered in Philadelphia did not arrive with a completed document in hand. They arrived with competing traditions of thought drawn from Montesquieu and Locke, from English common law and colonial charters, from their own lived experience of what had failed under the Articles, and the Constitution emerged not from any single authorβs vision but from the sustained collision and synthesis of those traditions over months of arduous deliberation.
The particulars of AI governance will be forged through a similarly deliberative and multi-disciplinary process. They will emerge from conferences where technologists, legal scholars, and policymakers sit in the same room and confront one anotherβs assumptions. They will be refined through deep engagement with academic literature in political science, institutional economics, and the history of how standards bodies have actually formed and functioned across industries and centuries. They will draw upon the rich body of theory that explains how societies construct shared norms under conditions of uncertainty, from the work of Elinor Ostrom on the governance of common-pool resources to the insights of constitutional political economy on the design of durable institutional frameworks. And they will be tested, revised, and tested again, because that is how every consequential standard in the history of organized human endeavor, from the Uniform Commercial Code to the Basel Accords to the Geneva Conventions, has come into being. No reader should mistake the absence of a finished blueprint for the absence of a path. The path is the deliberation itself, and the work of walking it has already begun.
The Stakes Before Us
There exists a persistent temptation, particularly within technology circles, to regard governance as a drag upon progress, as the bureaucratic weight that slows down the real and more exciting work of building. This view represents a fundamental misreading of history. The Constitution did not retard the growth of the American economy. It accelerated that growth by establishing predictable rules, enforceable contracts, and protected rights, thereby creating the conditions of confidence under which entrepreneurs, investors, and workers could operate with assurance that the ground beneath them would not shift without warning. Property rights enabled capital formation. Due process gave contract law its teeth. Freedom of expression sustained the marketplace of ideas upon which innovation itself depends.
The same logic applies with equal force to artificial intelligence. A credible governance framework does not strangle innovation. It produces the trust without which innovation cannot scale to its full potential. No enterprise will integrate autonomous AI agents into mission-critical operations if it cannot verify how those agents behave, what standards they uphold, and what remedies are available when something goes wrong. No consumer will place confidence in an AI-driven medical recommendation if there is no recourse when that recommendation proves harmful. No sovereign nation will permit cross-border AI commerce in the absence of a shared understanding of accountability and liability.
Without governance of a constitutional character, we are constructing a republic of intelligence upon sand. Agents will continue to proliferate. They will interact with one another and with billions of human beings. They will manage supply chains, adjudicate insurance claims, draft legal instruments, and allocate capital. And if they do all of this without common principles, without a shared and enforceable commitment to transparency, accountability, data rights, and interoperability, the result will not be the flourishing ecosystem that the optimists envision but rather a digital Articles of Confederation, fragmented and ungovernable and ultimately unsustainable.
The framers in Philadelphia did not enjoy the luxury of waiting until they understood every consequence of the system they were constructing. They acted upon imperfect knowledge, under enormous pressure, with the clear understanding that inaction was itself a choice, and a perilous one at that. We find ourselves in that very room today, and the question is not whether we shall write a constitution for the governance of artificial intelligence, but whether we shall write it deliberately, through inclusive and rigorous debate, or allow it to be written for us by default, by accident, and by whichever interests move fastest and lobby most effectively. The founders chose to write. So, I believe, should we.
AI-assisted insights, supplemented by 25 years of finance leadership experience.
