• integrated_82#Q1.4

RAIDT is novel because it reframes GenAI governance as an evidence problem at the point where risk actually materialises: the individual run. Instead of relying mainly on principle statements, model-level documentation, or periodic audit narratives, RAIDT sets the run as the unit of governance and requires a bounded run-level evidence pack for each material use. That evidence pack functions as a reconstructable proof object. It records what was asked, which model and tools were active, what context was retrieved, what output was produced, and what human or automated checks followed. In this sense, RAIDT moves governance from abstract assurance towards inspectable run-level evidence.

Its second novelty is measurement. RAIDT does not stop at preserving records; it converts the sufficiency of those records into a score profile across the five pillars (Responsibility, Auditability, Interpretability, Dependability, Traceability), using anchors 1=missing / 3=partial / 5=audit-ready. This makes governance readiness inspectable, comparable and improvable across runs, configurations, suppliers and domains, while preserving the dimension-level pattern rather than collapsing everything into a single opaque judgement. A further contribution is that RAIDT treats influence methods as governance interventions. Structured prompting, retrieval augmentation, PEFT or LoRA, and preference-based alignment are therefore not hidden engineering choices; they are logged conditions whose governance effects can be examined. The novelty of RAIDT is thus not merely that it documents use, but that it links reconstructable run evidence to a disciplined, comparable governance-readiness measurement approach.

Consider a public-service team using GenAI to interpret benefit eligibility rules. Without RAIDT, a staff member may keep only the final answer, so a later challenge cannot establish which rule version, retrieval result, or prompt structure shaped that advice. The output may sound confident, yet the organisation cannot reconstruct the basis of the decision.

With RAIDT, the same case yields a run-level evidence pack containing the prompt, model version, retrieval snapshot, document identifiers and hashes, plus any review step. The organisation can then assign a score profile across the five pillars (Responsibility, Auditability, Interpretability, Dependability, Traceability) using anchors 1=missing / 3=partial / 5=audit-ready. If retrieval augmentation improves evidence linkage, that becomes visible because influence methods as governance interventions are recorded rather than assumed. The novel value is that governance can now be inspected and compared at the level of the specific run, not inferred from policy claims alone.

• 08-RAIDT_Foundations_M_V50
• 11-RAIDT_Academic_Logic_M_v11