Sunset Provisions
Sunset provisions are regulatory instruments that build expiration dates into interventions, after which the regulation lapses unless the authorizing body, having reviewed evidence of the intervention's effects, actively renews it. The default is expiration rather than renewal, because the default of renewal produces institutional inertia — regulations that persist by bureaucratic momentum long after conditions justifying them have changed. The default of expiration forces the deliberative body to make an affirmative case for continuation, grounded in evidence about the intervention's actual effects, rather than allowing continuation to proceed by default. In AI governance, where technology evolves faster than any regulatory framework can anticipate and evidence accumulates too rapidly for static rules to remain adequate, sunset provisions are structural necessities rather than optional features.
In the AI Story
The interval for AI regulatory sunsets should be short — two to three years — reflecting the pace of technological change. An intervention designed for 2026 AI capabilities may be inadequate or excessive for 2028 capabilities. The sunset provision ensures the regulatory framework keeps pace with the technology it governs, not by anticipating changes that cannot be foreseen but by requiring periodic reassessment that evaluates appropriateness in light of changes that have actually occurred.
Sunset design interacts with the other three elements of AI regulatory architecture — disclosure, defaults, deliberation — in ways that produce functional architecture rather than collections of independent interventions. Disclosure creates the informational foundation on which defaults are designed. Defaults create the behavioral environment the deliberative body monitors. The deliberative body generates evidence-based assessments that inform revision of both disclosures and defaults at each sunset interval. The sunset provisions ensure the architecture remains responsive to evolving technology and accumulating evidence. Remove any element and the architecture degrades: without sunsets, the architecture becomes irrelevant; without deliberation, the architecture ossifies; without defaults, disclosure produces no behavioral effect; without disclosure, defaults are designed blind.
The political economy of sunset provisions is contested. Companies subject to regulation typically favor them as tools for rolling back interventions after political attention has shifted. Advocates for restriction typically oppose them for the same reason. Sunstein's position treats sunsets as neutral mechanisms whose political valence depends on the quality of the deliberative process that governs renewal decisions. A well-designed deliberative body, with adequate technical capacity and structural independence, uses sunsets to improve regulation over time. A captured or underresourced body uses sunsets to sanction deregulation regardless of evidence. The mechanism is only as good as the institutions that operate it.
Origin
Sunset provisions have a long history in American regulatory practice, dating to Colorado's 1976 adoption of comprehensive sunset laws for state agencies. Sunstein's treatment integrates sunset design into the broader framework of evidence-based regulation developed during his OIRA tenure and elaborated in subsequent scholarship on regulatory humility and institutional learning.
Key Ideas
Expiration as default forces affirmative review. When regulations expire automatically, continuation requires evidence-based justification rather than bureaucratic inertia.
Short intervals match rapid technology. Two-to-three-year sunsets reflect the pace of AI evolution; longer intervals produce regulations calibrated to capabilities that no longer exist.
Sunset design interacts with deliberation design. The mechanism's value depends on the quality of the body that conducts renewal review — sunsets are only as good as the institutions that operate them.
Political valence is contested. Regulated parties favor sunsets as deregulatory tools; advocates oppose them for the same reason; the mechanism itself is structurally neutral.
Appears in the Orange Pill Cycle
Further reading
- Sunstein, Cass, The Cost-Benefit Revolution (MIT, 2018)
- Gersen, Jacob, 'Temporary Legislation' University of Chicago Law Review 74 (2007)
- Sunstein, Cass, 'Regulation in an Uncertain World' National Academy of Sciences (2015)