Anam #02: "Atomic extraction has no grounding verification. Models hallucinate principles not present in the source." HIGH severity.
Problem: Stage 2 asks LLM to produce atomic principles with 16 fields. No verification between extraction and writing. Atomicity is not well-defined for LLMs. Three models produce structurally different outputs from the same paragraph.
Proposed: Stage 2b verification pass: for each extracted principle, require a verbatim anchor from the source chunk. If no anchor exists, flag confidence < 0.5. Add calibration procedure: labelled set of near-duplicate pairs, precision-recall curve, decision on acceptable false positive rate.
v3 2026-04-05 Q Mapped to whitepaper sections
v2 2026-04-05 Q Imported SPEC-039 from model_specifications_v2.html
v1 2026-04-05 Q Created spec: SPEC-039: Grounding Verification (Stage 2b)