Eugene Volokh is a First Amendment legal scholar and the founder of the legal blog The Volokh Conspiracy. He is the Thomas M. Siebel Senior Fellow at the Hoover Institution at Stanford (2024–present) and the Gary T. Schwartz Distinguished Professor of Law Emeritus at UCLA School of Law. In AI policy he is known for the User Sovereignty vs. Public Safety / Social Justice distinction set out in his contribution to the Digitalist Papers.
Background
Volokh's career has centered on First Amendment scholarship, particularly internet speech, trademark and expressive use, harassment law and speech-restrictive statutes, and employment speech. He has authored more than 50 law-review articles, and his UCLA First Amendment casebook is a standard reference. He is generally pro-speech across most contexts and characteristically opposes content-moderation-as-regulation, a libertarian-leaning posture rooted in distrust of concentrated editorial power whether exercised by private firms or leveraged by government.
He founded The Volokh Conspiracy in 2002 and continues as editor; the blog is now hosted at Reason magazine.
User Sovereignty vs. Public Safety / Social Justice Model
In "Generative AI and Political Power" (Digitalist Papers Vol. 1, September 2024), Volokh introduces the User Sovereignty vs. Public Safety / Social Justice Model distinction:
- User Sovereignty: tools that act as faithful servants of user intent (word processors, browsers, search).
- Public Safety / Social Justice: tools with creator-embedded guardrails that refuse outputs the creator deems harmful or immoral (ChatGPT, Claude, Gemini).
He argues that AI has shifted toward the latter, giving a small number of AI companies outsized influence over political life, and that this concentration is compounded by government pressure, echoing the concerns raised in Murthy v. Missouri.
Volokh offers five reasons for the shift:
- Authorship responsibility and legal liability.
- Feasibility, in that AI understands its own outputs.
- Guardrail creep, in which legal guardrails such as those for CSAM and libel beget ideological guardrails.
- Existential-risk framing that motivates pre-emptive content controls.
- A political zeitgeist in which the user-serving norm of the 1970s–90s was displaced by the equity-agent norm of the 2010s–20s.
His prescription combines competition, select structural government mandates, and First Amendment limits on government pressure, returning to user sovereignty as the default and applying targeted legal guardrails only where narrowly justified.
Relation to other positions
Volokh's AI essay is consistent with his broader skepticism of concentrated editorial power. His position diverges from several others tracked in adjacent pages:
- With Persily, who argues that targeted regulation is necessary, Volokh is skeptical.
- With Collective Constitutional AI, the positions run in opposite directions: CIP democratizes model values via public input, while Volokh wants models to defer to individual user intent.
- With Anthropic's safeguards approach (Source: anthropic.com), Anthropic explicitly rejects pure user sovereignty for frontier models.
Relationships
- supports: The Digitalist Papers (Stanford, Volumes 1–2) / User Sovereignty Model vs. Public Safety / Social Justice Model — his essay and concept.
- related: AI and the First Amendment / AI Political Bias / ChatGPT, Can You Solve the Content Moderation Dilemma? — adjacent concepts.
- related: Manhattan Institute — Measuring Political Preferences in AI Systems (Rozado) / Stanford GSB — Measuring Perceived Slant in LLMs (Westwood, Grimmer, Hall) — empirical evidence for Volokh's descriptive claims.
- related: Executive Order 14319 — Preventing Woke AI in the Federal Government — Trump EO that operationalizes Volokh-adjacent concerns.