•  
  •  
 

Abstract

Human-centered artificial intelligence (AI) is premised on the idea that humans must remain the ultimate locus of agency in technological systems. First Amendment jurisprudence has displaced this commitment. By extending speech protection to algorithmic architectures, the judiciary has forged a doctrinal shield that insulates corporate actors from the consequences of their products. Conversational AI deepens this crisis through intimate incursions into the epistemic and emotional registers of the self. Emerging approaches that anchor First Amendment coverage to the controllability or traceability of AI outputs offer no stable solution. Policy-level design choices such as safety guardrails and alignment processes can readily be recharacterized as editorial discretion, producing a doctrine whose scope fluctuates alongside engineering advances rather than constitutional principles.

This Article argues that the more productive inquiry focuses on the relationship between AI service providers and the publics they serve. First Amendment jurisprudence has never been indifferent to who claims its protection; across religious liberty, associational rights, and academic freedom, constitutional protection calibrates to specific governance structures and audience relationships. An institutional approach restores this nuance, distinguishing between ideological autonomy for conviction-driven institutions, associational autonomy for self-selecting communities, and organizational autonomy for those managing discourse without a particularized viewpoint.

Most conversational AI service providers fall into the third category. Within this framework, regulatory mandates for transparency and harm mitigation do not burden expressive freedom. Centering the cognitive freedom of the individual returns the doctrine to its foundational, human-centered origins.

Download

Share

COinS