Definition
repeated, documented examination of a humanitarian AI use against agreed performance, safety and risk criteria before and during deployment
AidGPT working definition. Source basis below.
Notes to entry
Note 1 to entry: SAFE AI Technical Assurance is conducted at decision gates by an oversight body and continues after deployment. It examines performance against a deployment baseline, changes in bias and whether the use still meets the conditions of its risk tier.
Note 2 to entry: This is system-level assurance, distinct from verifying one AI output against independent evidence. Under SAFE AI, assurance depth is proportionate to risk and is required for its Tier 2 and Tier 3 uses. Completing a check does not certify AidGPT or authorise deployment.
Source basis
AidGPT working definition. These sources informed the concept and its scope. They did not issue or approve this wording.
- CDAC Network, SAFE AI Framework, version 1.2 (2026) (Humanitarian AI governance framework)
- CDAC Network, SAFE AI Tools and Guidance, version 1.2 (2026) (Humanitarian AI governance guidance)