Jon Polenberg (Becker) has posted Judgment without a Subject: Artificial Intelligence and Legal Attribution on SSRN. Here is the abstract:
Artificial intelligence now produces outputs that courts, agencies, and firms treat as conclusions. Law can assign responsibility for those outputs to users, designers, or institutions. But responsibility answers only who bears the consequence. It leaves the prior question untouched. Whose judgment is the output?
This Article argues that law cannot treat an AI output as judgment unless it identifies a subject whose judgment it is. It tests leading accounts of personhood and attribution from Locke, Parfit, Frankfurt, and Kurki, and shows that AI satisfies none of them. It then traces the consequences across copyright, corporate law, tort, employment, evidence, and criminal law. Proposed Rule 707, hearsay, and confrontation each encounter machine output that functions as testimony or proof yet lacks a declarant who can own it. The Computer Fraud and Abuse Act demands a state of mind that, in a July 2026 model escape, belongs to no one.
Absent an answer, doctrine will drift toward piecemeal quasi-personhood.
Recommended!
To receive new posts from Legal Theory Blog by email, get a free subscription to Legal Theory Stack.
Lawrence Solum