Signal No One to Blame: a framework of constitutive AI unaccountability

Summary

This paper argues that prior research has framed AI accountability gaps merely as barriers that can be overcome through better standards or institutional reform. Against that view, the authors introduce the concept of constitutive AI unaccountability, describing configurations of actors, systems, and institutions in which accountability is conceptually impossible to establish no matter the effort applied. The study proceeds through three qualitative stages, a concept-centered literature analysis, a secondary analysis of interviews with 27 AI experts, and an illustrative application to the open-source agentic AI system OpenClaw. This produced nine categories and 20 themes spanning structural, technological, and normative clusters, along with eight interdependencies among them. The framework was turned into a 20-question diagnostic instrument, and applying it to OpenClaw detected 17 of the 20 conditions, including one configuration in which the AI agent was the only identifiable actor.