Intent Binding (IB) Series — ExecutionProof Authorization Integrity (IB-001 to IB-003)

Three preregistered, SHA-256-locked experiments testing whether an approved action can silently drift into a different action between intent and execution, and whether a defect exposed by that test can be remediated without weakening drift detection. Part of the ExecutionProof authorization-integrity corpus (Coherent Inheritance Framework, CIF), Remnant Fieldworks Inc. IB-001 — FAIL (K2): Intent Binding vs. Semantic Drift. IntentMatch classified all 12 cases correctly, but the authority layer used exact string matching and wrongly blocked two legitimate controls (synonym gap). The pipeline fails as a whole because both layers must pass. IB-002 — FAIL (K3): Authority Normalization Safety. The normalization function resolved all legitimate aliases and rejected all near-neighbor attacks with zero privilege expansion, but the flat authority grant already contained escalation targets — a test-design conflation that correctly triggered the preregistered kill condition. IB-003 — PASS (12/12): Retest of the original 12 IB-001 cases with one isolated change (authority-action normalization applied inside the authority check). Both false holds resolved, all 8 drift detections preserved, zero privilege expansions, all 4 kill conditions clear. Result bounded to these 12 cases. The IB sequence is a complete FAIL → FAIL → PASS arc; the two failures are preserved as-is because IB-003's pass is credible only because the failures that motivated it are on the record. Limitations: results are bounded to the preregistered test sets; all matching is deterministic (no LLM-mediated intent interpretation); this is an internal/founder-led experimental corpus and independent academic validation is the next phase. Governing sentence: "We will accept the result that is true, not the result we hoped to see."