AI-Authorship Accusations Against Human Discussants: A Pragma-Dialectical Account of Procedural Derailment

Abstract This paper offers a pragma-dialectical reconstruction of AI-authorship accusations directed at human discussants: the charge that a contribution is machine-generated or not attributable to a responsible participant. Such an accusation is best understood not as a factual label but as a meta-level procedural move. Introduced during the argumentation stage, it can shift the exchange from the merits of a standpoint to the discussant’s status and the admissibility of starting points, and under specified procedural consequences it triggers a nested opening-stage sub-discussion while leaving the original difference of opinion intact. The paper names the configuration in which discussant status, admissible commitments, and sufficiency thresholds are jointly reclassified procedural capture, and gives the conditions that mark it off from an ordinary personal attack or premise misuse. Its contribution is a symmetrical, default-governed repair protocol, with a warrant-filter for detector-based evidence, that restores critical testing of the original standpoint. The claim is normative rather than descriptive: the worked examples are analytic reconstructions.