Technical safety alone will not make AI rule accepted; law should convert performance into publicly recognisable authority by anchoring AI rule in recognized venues, publishing audience-facing reasons, and guaranteeing accessible review and remedies.

Main Finding

Legitimacy — understood as the public’s recognition that AI governance is rule‑bound, justified, and situated in institutions that speak for those governed — should be an explicit regulatory objective distinct from technical “alignment.” Legal design can structure recognition (turn performance into reasons and reasons into authority). To secure legitimacy, regulators should pursue three audience‑facing principles: Integration (locate AI rule‑setting in recognized venues), Familiarity (present rules and reasons in locally credible, intelligible forms), and Contestation (guarantee accessible, impartial review and effective redress).

Key Points

• Legitimacy ≠ Alignment: Accuracy, safety, or good performance do not by themselves confer public acceptance; legitimacy is an audience‑side sociological phenomenon (recognition), drawing on work in political sociology and procedural justice.
• Three core legitimacy failures for AI:
  • Opacity: Complexity and proprietary secrecy block comprehensible reasons, making decisions appear arbitrary even when correct.
  • Political integration / private power: Large private platforms and model providers exercise public‑facing authority (speech, visibility, access) without recognizable public authorship, creating de facto governance gaps.
  • Administrative automation: Agency use of AI can weaken notice, reason‑giving, participation, documentation, and review, undermining the state’s claim to rightful authority.
• Legal responses framed as thin vs thick legality:
  • Thin/formal legality: publication, stability, consistent application — signals non‑arbitrariness.
  • Thick/constitutional legality: public authorship, audience‑facing reasons, and contestability — supplies the forums, genres, and procedures that convert performance into recognized authority.
• Practical prescriptions (illustrative):
  • Seat consequential AI rule‑sets in venues recognized as authoritative (co‑regulation, public law constraints).
  • Publish system‑level principles, stable rulebooks, change logs, and plain‑language “reason letters” for high‑stakes outcomes.
  • Provide predictable second‑look channels: accessible appeals, impartial review, published timelines, and remedies that fix outcomes and rules.
  • Where per‑decision explainability is infeasible, use credible system‑level disclosures and effective contestation mechanisms.
• Limits of private proceduralism: Internal procedural features (appeals, oversight boards) can reduce arbitrariness but cannot replace public authorship or political community consent.

Data & Methods

• Methodological approach: conceptual and normative analysis drawing on political sociology (Beetham, Weber), procedural‑justice literature (Tyler), science & technology studies (Jasanoff), public‑law and administrative‑law scholarship, and empirical case studies of platform governance and agency automation.
• Sources and evidence: literature review and synthesis of scholarship and policy documents; illustrative examples include social media/search governance dynamics, Meta’s Oversight Board, administrative case studies, and U.S. policy instruments (e.g., OMB Memorandum M‑24‑10, ACUS guidance, Government by Algorithm survey).
• No original quantitative dataset or empirical experiment; argument rests on theoretical grounding, prior empirical studies of platform legitimacy, and documented agency practices.
• Analytical framing: identifies audience‑facing conditions (visibility of authorship, intelligible reasons, effective contestation) necessary for sociological legitimacy and maps legal/institutional levers to supply them.

Implications for AI Economics

• Demand and adoption: Legitimacy influences willingness to accept AI governance. Systems that are technically sound but illegitimate risk consumer withdrawal, boycotts, lower engagement, or political backlash — all of which reduce demand and platform/network value.
• Trust as economic capital: Legitimacy is a public good that increases cooperation and lowers transaction costs. Investing in audience‑facing transparency and contestation can raise long‑term trust and platform valuation, but generates up‑front costs.
• Compliance and operating costs: Mandates for public rulebooks, reason letters, documentation, logging, and appeals processing increase operational and compliance costs — particularly burdensome for smaller firms and startups, potentially raising barriers to entry and increasing concentration.
• Market structure and incumbency effects: Large incumbents may have advantages (resources to meet disclosure and contestation requirements, existing regulatory relationships). However, binding public constraints or co‑regulation can curb unilateral rule‑making power and alter firms’ competitive advantages.
• Incentives for product design and R&D: Shifting regulatory priorities from purely technical alignment to legitimacy‑oriented design reallocates firm incentives — more investment into explainability, user‑facing documentation, governance processes, and remediation mechanisms, possibly slowing feature rollout but increasing social acceptance.
• Emergence of complementary markets: Demand for third‑party auditors, independent reviewers, compliance platforms, logging/audit tools, and legal/governance services will grow. These create new economic opportunities and firms specializing in legitimacy infrastructure.
• Regulatory uncertainty and international competitiveness: Varied jurisdictional implementations of legitimacy rules (different civic epistemologies, venues of authorship) can fragment markets, increase compliance complexity, and affect where firms locate R&D and deployment. Harmonized approaches could reduce frictions; divergent regimes may advantage firms able to internalize cross‑jurisdictional costs.
• Policy design for efficiency and equity: Legitimacy interventions can raise social welfare if they prevent costly resistance, misinformation cascades, or systemic harms that arise from illegitimate governance. But they may also produce distributional effects (favor incumbents, raise consumer prices). Economic analysis should include recognition effects in welfare calculations and model legitimacy as a variable affecting adoption, network externalities, and policymaker/firms’ strategic behavior.
• Modeling recommendations: Incorporate legitimacy as an endogenous variable in models of platform competition, consumer choice, and regulator–firm games. Treat explainability, dispute resolution capacity, and public‑facing authorship as investments with returns in trust and reduced regulatory risk.

Short takeaway: Treating legitimacy as a regulatory objective changes the economics of AI governance — it imposes costs and creates markets for governance services, reshapes incentives and competition, and (if well designed) can increase durable value by converting technical performance into socially recognized authority.