Research & Publications

Redefining Freedom of Expression in the Age of Generative AI: Protecting Vulnerable Groups in Ephemeral Digital Spaces

Avigayil Finley, Elan Zashin, Danila Stepanov

This paper examines how First Amendment doctrine should respond to generative AI and ephemeral digital spaces, especially where AI-generated or mediated outputs function less like ordinary protected expression and more like conduct-like instrumentalities of exploitation, coercion, fraud, grooming, or accountability evasion. It focuses especially on minors, synthetic child sexual abuse material, synthetic identity harms, platform design, evidentiary collapse, and the limits of traditional "more speech" remedies.

Generative AI combined with ephemeral communication technologies challenges key assumptions of free speech doctrine: human speakers, stable artifacts, and functioning post hoc accountability. This paper argues for a bounded constitutional framework that preserves core First Amendment values while recognizing that some AI-mediated outputs and distribution architectures operate as instruments of exploitation, coercion, fraud, grooming, or evidentiary evasion. It proposes a four-factor "AI output as instrumentality" test and a three-lane regulatory architecture focused on high-culpability conduct, platform design duties, and procedural safeguards.