Use the existing governance records
Create one record for each defined use. Keep supporting evidence in the organisation’s existing governance documents and link to it from this record. The proposed AidGPT record organises the classification and references. It does not repeat an assessment or replace the SAFE AI Transparency Card.
| SAFE AI stage [F] | What the AidGPT record would connect | Where to keep the evidence |
|---|---|---|
| Problem and concept | Describe the use, alternatives, people affected, capabilities and initial assessment needs. | Use-case record and initial assessments. |
| Design | Record the task’s limits, data choices, community influence and proposed controls. | Design evidence and assessment updates. |
| Development | Link tests and supplier checks to the specific system version and intended use. | Test results, procurement records and evidence supporting assurance. |
| Deployment and review | Record what is authorised, monitoring, incidents, reassessment and plans to retire the system. | Transparency Card and linked operating records. |
| At each decision point | Record the responsible role, decision, conditions and what will trigger the next review. | Link from the AidGPT record to the original governance records. |
Use the invented examples for teaching. Before live use, complete the applicable assessments and authorisation process. Link the record to the relevant SAFE AI card sections: system purpose, humanitarian context, data and models, lifecycle, governance and references [F, p. 88]. Completing the record does not grant approval or establish compliance.
Source version
CDAC’s website dates SAFE AI version 1.2 to August 2026, while the archived framework cover says May 2026 [X, checked 25 September 2026]. Identify it by its version and digital object identifier (DOI). Authors: Helen McElhinney, Anjali Mazumder, Michael Tjalve, Suzy Madigan and Sarah Spencer. The archive confirms a CC BY 4.0 licence. Relevant pages were inspected in the browser preview, along with the public overviews and risk briefing.
The separate companion tools still need a field-by-field comparison before operational adoption.
Sources
F. SAFE AI Governance Framework