Synthetic Precedent: How Generative AI Reshapes the Citation Economy of Federal Courts

Citation is the currency of common-law reasoning: courts justify outcomes by locating them within a lattice of prior authority. This Article examines how generative AI systems, now routinely used by litigants and increasingly by chambers, alter the production and consumption of that authority. Drawing on a hand-collected dataset of 4,812 federal district court opinions issued between 2021 and 2025 that discuss or sanction AI-assisted filings, the Article documents three shifts. First, a measurable rise in citation density in briefs correlates with AI drafting assistance, without a corresponding rise in citation accuracy. Second, courts have begun to develop what I call verification doctrine — procedural rules allocating the burden of confirming that cited authority exists and stands for the proposition asserted. Third, the citation economy is becoming reflexive: AI systems trained on judicial text now influence which precedents are surfaced, cited, and thereby reinforced, creating feedback loops that entrench some lines of authority and orphan others. The Article argues that verification doctrine should be understood as a species of evidence law applied to legal argument itself, and proposes a disclosure-and-spot-check regime that preserves the efficiency gains of AI drafting while protecting the integrity of precedent.