He on a Hohfeldian Analysis of AI Governance

Mingdong He has posted Floors and Ceilings: A Hohfeldian Analysis of Artificial Intelligence Governance on SSRN.  Here is the abstract:

Debate over artificial intelligence governance is organised along a single axis running from permissive to restrictive. This article argues that the axis obscures a distinction of jural type. Using Hohfeld’s analysis of fundamental legal conceptions, I show that instruments imposing determinate obligations (floors) and instruments licensing persons (ceilings) operate at different orders of legal relation: the first imposes a duty correlative to a right; the second confers a power on an administrative body, correlative to a liability in the researcher — a susceptibility to having one’s legal position altered without having done anything wrong. The second-order character of ceilings explains four otherwise puzzling features: their resistance to textual constraint, their attractiveness to capture, the monotonicity of their incidence in firm size, and their resistance to downward revision. I derive a six-part test classifying a proposed instrument from the face of its text — distinguishing floors, defective floors, and ceilings — and apply it to four near-simultaneous interventions of 2025–2026, clause by clause to the EU Artificial Intelligence Act — which contains both forms — and to the Chinese position, whose pairing of domestic permission with state-encouraged diffusion both instantiates the framework’s predictions and disciplines its addressees elsewhere. The claim that ceilings redistribute participation toward incumbents is evidenced from the GDPR natural experiment and the occupational-licensing record. Comparative analysis of recombinant DNA governance, pharmaceutical authorisation, and aviation incident reporting identifies when a ceiling is warranted; eight floor-compatible instruments are set out, with a candid feasibility note on mandatory insurance. I conclude with the conditions on which the framework’s fate depends, including the one that would defeat it on the point that matters most.

Recommended!

To receive new posts from Legal Theory Blog by email, get a free subscription to Legal Theory Stack.

Lawrence Solum