The Blind Sniper: AI, Epistemic Authority, and the Conversion of Probability into Condemnation

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

This paper examines the institutional mechanism through which probabilistic outputs from AI detection tools are converted into binding punitive decisions. Using the 2026 Thélyson Orélien / Académie Goncourt controversy as a starting point, the study introduces two original conceptual frameworks: the Detection-to-Sanction Gap (the inferential collapse between detection, attribution, misconduct, and sanction) and the Epistemic Authority Gap (the distance between an algorithm's statistical capacity and the institutional authority granted to its output). The author argues that the primary threat is not merely algorithmic error, but the delegation of institutional judgment to probabilistic signals without adequate procedural safeguards. The paper extends this analysis from literary prizes to academic integrity in higher education and risk assessment in criminal justice, proposing a normative framework based on clear evidentiary thresholds, burden-of-proof allocation, and procedural fairness.