Parametric memory auditing probes what language models recall unprompted from training data, while grounded memory auditing measures live search-derived facts to expose hallucinations and coverage gaps via the Fact & Gap Checker.
flowchart LR
A[Brand Topic] --> B[Parametric Probe: No Search Tools]
A --> C[Grounded Probe: Web Search and URL Context]
B --> D[Raw Internal Recall]
C --> E[Live Source Facts]
D --> F[Fact and Gap Checker]
E --> F
F --> G[Confirmed / Contradicted / Gap Status]Referenced by