AI Change Accountability & Reauthorisation
Operational governance for AI change after approval.
Method edition:
AI Change Accountability & Reauthorisation (ACAR) is a procedural method for governing changes to AI systems and AI-enabled workflows already approved or in use. A different model, a new provider, expanded permissions or reduced human oversight can alter the basis of an earlier approval. ACAR connects the review of change with organisational authority, implementation responsibilities and evidence that the resulting decision has taken effect.
How ACAR works
ACAR examines one proposed or observed change, or risk signal, in relation to the relevant baseline and the status of any prior approval. The materiality review considers implications for affected people, human oversight, decision authority and external dependencies. It uses structured judgement and defined escalation rules to determine the review route: documentation only, enhanced monitoring, specialist reassessment or formal reauthorisation.
ACAR separates three decisions: the review required, permission to operate during review and the final organisational outcome. Formal responsibility and actual decision authority are recorded separately; authorisation requires both a valid mandate and the practical power to require implementation, impose conditions or stop use.
The decision sets the applicable closure criteria and any required actions, owners, deadlines and verification responsibilities. Closure requires evidence that those criteria have been met. The completed record connects the original change, the decision and its rationale with implementation evidence, continuing obligations and the grounds for later review or reopening.
Applying ACAR
ACAR is intended for AI and product owners, business and process owners, governance leads and authorised decision-makers. It can be used for live changes, retrospective reviews and tabletop exercises, or embedded in a standing organisational process.
Organisations can apply the method internally or with a facilitator. The guide includes a readiness check for a defined operating scope, with named ownership and access to qualified specialist input where needed. Before a decision or closure, the authority and evidence needed for that step must be confirmed. Intake and urgent protective action can proceed while readiness gaps are addressed.
ACAR complements existing change, incident and risk management by addressing the authority and approval questions that a technical change process may leave unresolved. The method is platform-neutral, and its records can be mapped into existing organisational systems, with one designated system of record preserving the separate decisions, evidence references and history.
In ACAR, reauthorisation means a renewed organisational decision about whether, and under which conditions, an AI use may continue. Legal advice, technical model evaluation and conformity assessment remain separate activities. ACAR does not confer regulatory authorisation, certify an AI system or establish its compliance.
Download the method
This edition combines the reference guide revised on 29 October 2026 with the compatible supporting materials dated 30 September 2026.
- Start Here and workflow — An introduction to the package and a visual guide to the process. Start Here (PDF) · Workflow (PDF) · Workflow text (MD)
- Reference guide — Scope, materiality and escalation rules, decision requirements and implementation guidance. PDF · DOCX
- Working templates — The Organisation Authority Map, two-part AI Change Event Sheet and supporting records for closure and external handoffs, with a short form for eligible routine changes. DOCX · PDF
- Event register — A workbook for tracking events, actions, open issues and decision history. XLSX
- Worked examples — Two fictional cases, with completed forms and a matching example register. Examples (PDF) · Examples (DOCX) · Example register (XLSX)
Download the complete package (ZIP)
The materials are free to download, use and adapt, including commercially, under Creative Commons Attribution 4.0 International (CC BY 4.0). Credit the author, link to the licence and indicate any changes. Completed organisational records do not have to be made public.
Developed and maintained by Maja Kurek, RuleBridge Advisory.
Updates and corrections are recorded in the changelog; earlier editions remain available. For questions or feedback about the method, contact office@rulebridge.eu.