AI Transparency: Governance Compliance or Stakeholder Requirements

AI Transparency: Governance Compliance or Stakeholder Requirements? This repository contains the supplementary materials accompanying the accepted RE 2026 (https://conf.researchr.org/home/RE-2026) paper AI Transparency: Governance Compliance or Stakeholder Requirements?. The materials provide details to the evaluation instrument, scoring guidance, stakeholder mappings, dataset, and supporting analysis artefacts. The resources enable replication of the study, secondary analysis of the dataset, and comparative investigations of AI transparency practices across jurisdictions and governance contexts. ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━FILES INCLUDED━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1. AI_Transparency_Statements_URLs.csv Full list of all 100 Australian Government agencies within scope of the DTA mandate, including agency name, URL to the published AI transparency statement (where available), and public contact details. Establishes corpus traceability and enables independent verification of all analysed statements. 2. AI_Transparency_Evaluation_Rubric.pdf The complete 20-criterion evaluation instrument used to assess transparency statements. Includes scoring scale definitions (0–2), criterion descriptions and evaluation questions, thematic grouping across three criterion groups, stakeholder class mapping within the RCIN framework, and supplementary qualitative metrics (Ambiguity Score; Sentiment Tone). See Policy Versioning Note below. 3. RCIN_Framework_Diagram.pdf Visual representation of the RCIN (Risk, Control, Involvement, Need) framework as a practical four-stage methodology for evaluating AI transparency artefacts against stakeholder-specific needs. Illustrates how the framework can be applied by any organisation or regulatory body operating under an AI transparency mandate, independent of the Australian context. 4. Supplementary_Scoring_Results.pdf Full empirical results for all 92 evaluated agencies, including: summary statistics, per-criterion score distributions, thematic group comparisons, governance-oriented versus contestability-oriented criterion analysis, complete criterion-level scoring table for all agencies, qualitative evaluator observations and criterion-level notes (76 of 92 agencies), and the list of 8 agencies without published statements at the time of data collection. ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━POLICY VERSIONING NOTE━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ Scoring was conducted against Version 1.1 of the DTA Policy for Responsible Use of AI in Government (effective 1 September 2024). A revised Version 2.0 took effect on 15 December 2025 and introduces additional mandatory requirements including strategic AI adoption planning, use case accountability, and risk-based impact assessment.The current standard is available at: https://www.digital.gov.au/ai/ai-in-government-policy/standard-ai-transparency-statements Replication against Version 2.0 may yield different results. Researchers extending this study should verify the policy version applicable at the time of analysis. ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━DATA COLLECTION━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ Transparency statements were collected and evaluated in May 2025. Statement availability reflects the state of agency publications at that time. Agencies that have since published or updated statements are not captured in this dataset. ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━CONTACT━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ For methodological questions or replication inquiries, please contact the corresponding author via the associated publication.