Rule of Law

The stakes couldn’t be higher — and the time is now. As AI agents perform more societal work, our choice — govern them by the rule of law, or let them run unchecked — will set the stage for the near-term and long-term futures, for better or worse. In this moment, we must recommit to the rule of law. I’ve spent a career making the law something people — and now machines — can rely on.

The bench — the rule of law, applied without bias

I began my career clerking for two chief judges — one on a state appellate court, one on a federal district court. There I helped apply the rule of law as it must be: faithfully, on the record, and regardless of any judge’s personal preference. A ruling must follow from the law and the facts — not from proclivities or whims. That’s the rule of law in its purest form, and it has set the standard for everything I’ve done since.

The bar — the rule of law, in defense of rights

For more than a decade as a complex-litigation lawyer — and later as a cybersecurity and digital-forensics lead — I invoked the rule of law to uphold my clients’ rights. Litigation is where abstract principles become concrete protection: contracts enforced, rights vindicated, wrongs made whole. Arguing the law daily taught me that the rule of law is only as strong as our ability to find it, read it correctly, and apply it consistently. And justice delayed is justice denied.

The corpus — the largest body of law on Earth

I then sought out the largest corpus of law I could find — every case, statute, and regulation — ultimately helping build systems that computationally analyze the laws of more than 100 countries. Exploring that entire legal corpus changes one’s perspective: law isn’t a static rulebook but a vast, living, contradictory, constantly-changing body of written human values. EU values differ from Texas values; both are encoded in their law. To serve the rule of law at that scale, and for agents to be able to effectively parse those human values, the law itself must become legible to machines.

Parsing the law — making it computable, so it can be enforced

For the last ten years, my work has been to parse that law — to make it computable. With the open-source SALI / FOLIO standard (which I led for five years, growing it from ~1,000 to 18,000+ tags), the law becomes a deterministic, queryable knowledge graph of rights, obligations, and prohibitions. The point is not novelty for its own sake; it is enforcement. When the law is computable, it can be better interpreted and more faithfully applied — by lawyers, by in-house and corporate counsel, by arbitrators, and by the courts themselves.

Doing this well means tracking the law as it moves: which holdings have been overturned, which statutes amended, how new law evolves on top of old. A decade of that work gives me a rare, structural view of law as a living system — and a tested ability to parse it, interpret it, and communicate why the rule of law matters to the very stakeholders charged with disseminating and enforcing it.

Standing up for the rule of law

The rule of law isn’t merely my vocation; it’s a conviction. When constitutional norms came under strain in January 2026, I drafted and organized an open letter supporting the rule of law — signed by more than 650 lawyers, judges, and legal technologists, across geographic and political lines. It opens with a simple affirmation: “We support the rule of law.” It then says what’s at stake for everyone who works in or relies on the law (that’s everyone):

If the rule of law is ignored, then legal advice loses meaning. Legal tech loses value. If the powerful can disregard the law with impunity, then our legal analyses, compliance programs, and contract drafting become performative exercises: accurate but irrelevant.

Democracies don’t legitimate force by outcome; democracies legitimate power by process, principles, and accountability. … Without the rule of law, businesses lose value. And the world becomes poorer.

We closed with the conviction I hold most deeply: “Because without the rule of law, our lives — our work, our institutions, and our society — all lose value.” That’s the same conviction behind everything above — and the reason I want to help build AI that is legal by design. That conviction is even more urgent now than when we wrote it: as AI scales, the moment to anchor it to the rule of law is now.

Why it matters now — AI and democracy

The rule of law is what keeps a democracy alive: democracies legitimate power not by outcome or by force, but by process, principles, and accountability. When that erodes, democracies die — and AI is now part of the equation. Without conscious direction, AI can tilt toward concentrated, unchecked power: surveillance at scale, automated decisions with no recourse, authority without accountability. With conscious direction toward the rule of law, it does the opposite — making rights legible, holding power accountable, and putting the law within reach of everyone it governs.

As agents perform more of the world’s tasks, we have to choose how they act: laissez-faire, anything-goes — or constrained by rules. Two rule-sets matter most: law (whatever a given jurisdiction requires) and ethics (the values of the user and those affected). What we build into these systems now will set the stage for years to come. The stakes couldn’t be higher, and the time is now — we must not squander this moment.

The same logic governs markets: capital trusts stable rules, not political whims — so an economy, like a democracy, is only as strong as the rule of law beneath it.

That direction is what I’ve been building: AI that is legal by design — aligned not to “human values” in the abstract, but to the laws that encode them and the democratic processes that legitimize them. Capable AI systems, acting across jurisdictions, will routinely brush against the law — and each law encodes a jurisdiction’s human values — so aligning AI to human values includes aligning it to human law. I’ve defended the rule of law in public — drafting and organizing a rule-of-law letter signed across political lines — and spent a decade making law computable. The Anthropic Institute’s “AI & Rule of Law” mission is where those two halves meet.