Bridget C.E. Dooling (Ohio State University (OSU) – Michael E. Moritz College of Law) has posted Outside Helpers, Rulemaking, and the Presumption of Regularity (Theoretical Inquiries in Law, forthcoming) on SSRN. Here is the abstract:
It is no longer hypothetical: U.S. government agencies are using AI to write regulations. We therefore must grapple with the intersection of the Administrative Procedure Act (APA) and the undetectability of AI-generated material. To challenge an agency rule for improper agency use of AI under an APA theory, a litigant would need to know which regulations were drafted using AI. But agencies do not acknowledge their use of AI in specific rulemakings and AI-generated text can pass undetected, leaving would-be litigants with an information gap. There are ways in which AI and other outside helpers can legitimately aid the rulemaking process, but policy decisions are human terrain. The APA and current doctrine may not be equipped to hold the line if AI-written preambles can hide in plain sight.
As it stands, the public does not know who writes what in a regulation; we don’t usually know, for example, which staff members worked on any given rule, or whether regulatory language was supplied by outside interests. This opacity—reinforced by the presumption of regularity—might have been acceptable when we could reasonably assume that agencies were in control of writing their rules. Agency use of contractors to write rules has been chipping away at this assumption, but AI undermines it completely. We don’t know who is writing rules anymore because of the availability of algorithmic tools, wide-eyed institutional enthusiasm for AI, and ongoing political hostility towards proceduralism, the civil service, and other good government bulwarks.
Recommended!
To receive new posts from Legal Theory Blog by email, get a free subscription to Legal Theory Stack.
Lawrence Solum