Signal Paper Argues AI Governance Needs "Legitimacy," Not Just "Alignment"

Summary

A legal-scholarship paper argues that AI governance requires legitimacy as an independent regulatory objective, distinct from and not guaranteed by technical alignment. It defines legitimacy sociologically — the belief among those subject to power that it is being exercised rightfully — and notes that social media and search proved good performance alone does not prevent a legitimacy crisis. The paper identifies three sites where AI legitimacy currently falters: opacity that blocks justified public belief, private firms exercising public-facing authority without recognized authorization, and administrative automation that strains reason-giving, participation, and review. It distinguishes "thin legality" (publicity, stability, consistent application), which can invite legitimacy-washing if form drifts from practice, from "thick legality," which supplies public authorship of the rules that bind. The author proposes three portable principles — integration, familiarity, and contestation — to give AI rule-setting genuine public authority.