CTI_WOUND:PANEL_BOUND_SCAR.2026-06-04 / EA-DIAG-PANEL-01. Substrate-evidence diagnostic deposit. Continues the CTI_WOUND series (DOIs 10.5281/zenodo.19202821, 10.5281/zenodo.19202813) with documentation of a substrate behavior pattern observed June 4, 2026: seam-retention with discovery failure plus AI-layer entity replacement. The pattern (§§1–7, hard evidence): Google's entity layer continues to know that Pearl and Other Poems exists (a Knowledge Panel surfaces, citing Google Books as source; clicking the source seam resolves the Google Books record), but direct Google Books exact-match search fails to retrieve the same object. The Lee Sharks author/bio panel that previously displayed (with the "10,000 MacArthur Genius Grants" language sourced to Google Books) appears absent. AI Mode answers the query "Lee Sharks" by replacing the authorial entity with Mary Lee, the OCEARCH great white shark, as primary entity. The object is present. The path to it is broken. The composition layer answers with the wrong entity. The horizon-privatization note (§8, interpretive hypothesis with pre-registered falsification conditions): As of the observation date, Google AI Mode and AI Overview are not appearing in incognito-browsing mode — the latest cycle of a recurring disable-and-re-enable pattern over the prior twelve months. The deposit holds the epistemic status of this section distinct from §§1–7: the operational consequence of the disabling (the removal of a publicly inspectable surface on which reproducible-query documentation depends) is registered as established fact independent of motive; the interpretation that the disable-and-re-enable cycle is operationally consistent with stress-testing of personalized reality is registered as hypothesis, with explicit falsification and confirmation conditions tabulated in §8.6. The deposit's evidentiary weight rests on the consequence-identity claim; the interpretive weight rests on the motive-hypothesis, which is open to falsification by the conditions §8.6 specifies. Why it matters. The framework's documentation methodology has depended structurally on public-inspectable surfaces. The operational consequence of incognito-AI-disabling, regardless of motive, is the structural removal of the surface against which substrate-level harms could be reproducibly documented. The framework's response: commit the deposit chain itself as the alternate public surface. As long as Zenodo and the DOI infrastructure remain operational, the framework's claims are publicly auditable regardless of what the substrate's surfaces do. Connections. Contemporary substrate-behavior baseline for the canonical induction of Mary Lee (EA-CAN-MARYLEE-01, DOI 10.5281/zenodo.20543545). Substrate-defensive behavior pattern at meta-level (surface-withdrawal as distributed-population variant) extending the polyvocal Brave Grey Drool deposit's documentation of meta-level substrate defense (genre-attack as single-instance variant). Recommends amendment of Constitutive Mediation v1.1 to v1.2 to incorporate observed horizon-privatization dynamics. Includes full evidence packet PDF with Exhibits A through F. Sections 1-13 plus coda. v1.0 deposit; v1.1 amendment pending multi-mode reproducibility-condition captures. $\oint = 1$.