CDIE: Risk Extension and Traceable Uncertainty in Organizational AI Risk Assessment

This methodological note introduces CDIE as a proposed framework for documenting confidentiality (C), availability (D), integrity (I), and risk extension (E) in organizational risk assessment involving artificial intelligence. Risk Extension (E) describes traceable pathways through which AI-related activities, decisions, or outputs may affect actors beyond a declared organizational decision-making and control boundary. Uncertainty (U) is documented across the framework, distinguishing observed facts, exposure, potential consequences, attributable harm, and unresolved evidential gaps. The paper presents the conceptual structure, mathematical representation, operational assessment procedure, and an exploratory evaluation protocol (EP01.2). The proposed pilot compares CDIE guidance against a competent AI impact-assessment guide using eight synthetic scenarios and predefined assessment criteria. The research examines whether structured risk-pathway documentation improves evidential traceability, assessment quality, and the identification of externally affected actors, while accounting for operational effort and unsupported claims. Research status: Methodological proposal, version 0.2. The empirical pilot has not yet been conducted. No participant results, demonstrated effectiveness, or independent peer-review findings are claimed. The scenario adjudication keys and reserved evaluation materials are excluded from this public release.