Damien Riehl — application for Member of Staff, AI & Rule of Law at Anthropic

01 ⁄ 04

AI safety evaluations · a legal-alignment lens

“…developing frameworks to assess whether frontier AI systems behave consistently with constitutional norms, due process, and rule-of-law principles.”

What's safe is what's legal — and I built the standard (and OWL knowledge graph) that defines it.

After working for the world's largest legal-technology company, Thomson Reuters — used by the world's largest firms and corporations — I drew on that experience to lead SALI, the legal-data standard, for five years: growing it from ~1,000 tags to 18,000+ and assembling a coalition of the world's largest law firms, the largest legal-tech companies, and the largest corporations (Microsoft among them) to standardize the metadata that classifies everything that matters to the business and the practice of law — worldwide (UK, EU, India, South America, APAC).

The question SALI and FOLIO answer is precisely what's safe — that is, what's legal? They track human values through the lens of legal values: “do not murder” and “do not steal” are at once ethical, moral, and legal — and Murder and Theft sit among the 18,000 tags, translated across modern languages into a knowledge graph of legal and normative human values.

02 ⁄ 04

Institutional vulnerability analysis

“…identifying where AI creates novel pressure points in democratic governance, and proposing both policy and technical mitigations.”

Legal by design — because the alternative can break democracy.

When a Stanford professor's expert declaration was caught citing AI-hallucinated cases, a federal court struck it — and quoted my writing on exactly that danger.

My work helps systems be legal by design — following both governmental law and ethical and moral principles. Built that way, these systems can do enormous good in the world. Built the other way — imbued with hidden or nefarious principles — they can be devastating to democracy. It matters enormously that we get this right.

As both a litigator and a systems builder, I can build the technical systems and the legal systems together, so each reinforces democracy rather than erodes it. Legal by design. Preventive law.

03 ⁄ 04

Novel legal issues in frontier AI

“…mapping the unsettled legal terrain created by capable AI systems, including questions of liability, agency, and institutional authority.”

Mapping the unsettled law of capable machines.

My 25-year career has lived where the law has no settled answer — autonomous-vehicle liability and agency (a direct analog to AI agents), digital first-sale and ownership, and the copyrightability of machine-made work. Two issued patents; three trial-practice books.

With All the Music, my team brute-forced every melody that can exist — all 471 billion of them — and released them to the public domain (CC0) to defuse “you stole my melody” lawsuits.

04 ⁄ 04

Applications that bolster democratic processes

“…developing or identifying ways AI could strengthen civic participation, legislative capacity, or judicial function.”

My entire career has bolstered democracy — by scaling justice.

The through-line of everything above is access to justice. I build coalitions (SALI, FOLIO). I work with the world's highest courts (supreme courts nationwide). I serve access to justice (legal-aid organizations worldwide). In my current role at Clio I help build Vincent for a 110+-country dataset — and that same experience can help Anthropic's products and services further access to justice. Justice must be scaled — and I can help Anthropic scale it.