Regulatory managerialism is a term used by the legal scholars Julie E. Cohen and Ari Ezra Waldman to describe the importation of private-sector managerial practices — and the ideologies embedded in them — into the activity of public regulation. The framing was set out in their introduction to a 2023 symposium issue of Law & Contemporary Problems, "Introduction: Framing Regulatory Managerialism as an Object of Study and Strategic Displacement," 86 Law & Contemporary Problems i (2023) (Source: https://scholarship.law.duke.edu/cgi/viewcontent.cgi?article=5116&context=lcp).
Background
In Cohen and Waldman's account, regulatory managerialism names a mode of governance in which regulators adopt the tools, metrics, and assumptions of corporate management — process documentation, risk registers, internal compliance functions, auditing, and standardized benchmarks — as the substance of oversight. The authors treat this as an object of study and discuss strategies for "strategic displacement," that is, for reorienting or broadening the approach so that managerial techniques do not substitute for the underlying public ends of regulation (Source: https://scholarship.law.duke.edu/cgi/viewcontent.cgi?article=5116&context=lcp). The term has been taken up in adjacent legal scholarship on digital and platform regulation.
Use in AI governance
In AI-policy debates the term is invoked as a critique of how AI safety institutes and risk-management frameworks operationalize oversight. The critique holds that combining public regulatory authority with deference to private-sector practice can produce governance that is procedurally elaborate but narrow in what it actually checks. Scholars in the Deirdre K. Mulligan cluster of sociotechnical-governance work argue that bodies such as CAISI (formerly the US AI Safety Institute, NIST CAISI (Center for AI Standards and Innovation)) and the UK AI Safety Institute (UK AI Safety Institute (AI Security Institute)) have leaned toward model-centric, technocratic evaluations — relying on industry-standard benchmarks and red-teaming — despite framings such as the NIST AI Risk Management Framework 1.0 that define risk management as coordinated activities to direct and control an organization rather than only a model. In that argument, treating benchmark performance and red-team exercises as guarantees of safety, rather than as discovery of potential weaknesses, exemplifies regulatory managerialism applied to AI.
Debates and positions
The concept is used critically rather than as a neutral description, and its application to AI safety institutes is contested. Proponents of evaluation-based oversight argue that standardized benchmarks and red-teaming provide tractable, comparable, and improvable measures in a field where few alternatives exist; the managerialism critique responds that such measures can displace attention from organizational, distributional, and sociotechnical dimensions of risk. The dispute connects to broader debates over whether AI risk management should be model-centered or organization- and deployment-centered.
Relationships
- related: Safetywashing — an adjacent critique of compliance-as-substitute-for-safety.
- related: Ai Risk Management (stub), NIST AI Risk Management Framework 1.0 — the risk-management framing the critique engages.
- related: Deirdre K. Mulligan — associated with the sociotechnical-governance scholarship that applies the term to AI safety institutes.
- related: NIST CAISI (Center for AI Standards and Innovation), UK AI Safety Institute (AI Security Institute) — institutions named in the AI-governance critique.
Sources
Page created 2026-06-14 (gap-scan) from the canonical Cohen & Waldman (2023) introduction on the Duke Law scholarship repository. The full Law & Contemporary Problems symposium issue is a foundational ingest candidate to deepen this page.