Work

Hidden Support and Causal Dependency Depth: Measuring Operational Autonomy in Artificial Agents

Year: 2026

Type: preprint

Abstract: Artificial agents can perform tasks successfully while depending on environmental resources or infrastructure maintained externally. This work operationalizes hidden support dependence through Support Withdrawal, Autonomous Dependency Management (ADM), and Causal Dependency Depth (CDD). A fully traceable simulation suite tests support withdrawal, task-only instrumental maintenance, and depth-indexed causal-vs-decoy intervention specificity. The package includes frozen protocols, source code, deterministic seed rules, raw data, statistical outputs, figures, supplementary methods, prior-art positioning, and SHA-256 integrity records. The results concern operational dependency regulation and do not establish consciousness, intrinsic self-preservation, or deployment-level autonomy in contemporary AI systems.

Source: Zenodo (CERN European Organization for Nuclear Research)

DOI: 10.5281/zenodo.23300035

Author: Manco, Alberto

Language: English

Topic: Ethics and Social Impacts of AI

Subfield: Safety Research

Field: Social Sciences

Domain: Social Sciences

Open Access status: green

OpenAlex ID: W7221457096

Cites: 2

Cited by: 1

Related to: