We are a partnership of former regulators, model laboratory veterans, and academic lawyers, working at the intersection of AI capability and the rules that govern it.
The firm was founded in late 2021, in a year when the European Commission's first AI Act proposal was still being read more as a discussion document than as a binding text, and when the idea of a general-purpose model laboratory facing systemic risk obligations would have struck most lawyers as a category error. Our founding partners had spent the previous decade working on data protection, platform regulation, and dual-use technology controls, and had concluded that those regimes, while instructive, would prove insufficient as templates. A new practice was needed.
Five years on, the bet has held. The legal regimes that govern frontier AI — the EU AI Act and its Code of Practice, the patchwork of U.S. state transparency statutes, the Commerce Department's diffusion framework and dual-use foundation model rules, the General Court and U.S. district court rulings on training data and fair use — are increasingly treated as a coherent body of law. We helped clients navigate that body of law as it formed, and we continue to advise as it matures.
Thule is a partnership of fifteen practitioners and a small group of senior advisors. We deliberately do not publish individual biographies. Our work for sovereign clients, model developers, and listed deployers depends on a degree of institutional discretion that public profiles tend to erode, and our partners have asked, with some firmness, that the firm's website not become a directory of their previous affiliations.
We can, however, describe the team in aggregate. Among our current partners are former officials of three competition authorities (one G7, two European), two senior trade negotiators with experience at the OECD and the WTO, a former member of a national constitutional court, two veterans of policy and safety functions inside frontier model laboratories, and an academic lawyer who has held a chair in the law of artificial intelligence at a leading European university. Several partners hold dual qualifications across common-law and civil-law jurisdictions. All have held front-line policy or counsel roles during the period in which the AI regulatory regime has been built — not merely commented on it from outside.
Beneath the partnership is a small group of associates and research fellows. Our policy is that every engagement is led, in fact and not in name, by a partner. We do not staff engagements with junior teams supervised at a distance, and we decline mandates whose volume would force us to do so. This is, we recognize, an inefficient model. It is the model our clients pay for.
In the four years since our founding, the firm has acted on mandates spanning the EU AI Act's negotiation and implementation phase, the General-Purpose AI Code of Practice drafting rounds, the U.K. and U.S. AI Safety Institute (now Security Institute) evaluation programs, the Bletchley–Seoul–Paris summit track, the post-rescission reorganization of U.S. federal AI authority under the AI Action Plan, multiple state-level legislative interventions including California's SB 1047 and SB 53, and a number of copyright and consumer-protection matters that we are not at liberty to enumerate.
We have testified before legislative committees on three continents, contributed to several public consultations including the AI Office's series of GPAI rulemakings, and published amicus submissions in matters before U.S. federal courts. Our partners have been seconded, in various individual capacities, to the secretariats of the OECD AI working party and the G7 Hiroshima Process.
Thule is wholly owned by its partners. We accept no equity from clients, hold no positions in the public securities of the foundation model developers we advise, and do not staff engagements in which a partner has a personal financial interest in the outcome. We accept that this discipline costs us mandates. We believe the trade is worth making.
We also publish our views, in the Viewpoints section of this site, even when those views are inconvenient to clients we have served. Our advisors retain the right to disagree publicly with positions the firm has taken in regulatory submissions, and several of them have exercised that right. We regard the discipline of writing for the public as one of the conditions for being a competent advisor in private.
New mandates are accepted by referral or after a preliminary call with a partner. We are happy to discuss potential engagements in confidence before a formal retainer is proposed, and our standard practice is to provide a written scope and fee estimate within ten working days of a first conversation.
For inquiries: counsel@thule-advisors.example. For press and academic correspondence: press@thule-advisors.example.