What replaces a redacted field in the court-facing layer?

Each protected field is replaced in place, at its original position in the document structure, with a placeholder such as [REDACTED-PHI sha256: ] - a truncated hash of that specific field's original value.

How is this different from typical PDF or e-discovery redaction?

Conventional redaction tools operate at the document or block level and offer no cryptographic relationship between the redacted output and the original - a reader must simply trust the redaction was done correctly. JIL's per-field commitment lets any single redacted field be independently, cryptographically verified against the sealed original.

Does the sealed unredacted layer ever leave the system?

The claim describes generation of both layers from the same source records with restricted access to the unredacted layer; access-control specifics are an implementation detail outside the scope of this claim, which covers the redaction and verification mechanism itself.

Per-Field Inline Cryptographic Redaction Commitments

Q: How does this prove the redacted document is faithful to the sealed original?

If a specific field's original value is later disclosed, a verifier can recompute its hash and confirm the truncated prefix matches the placeholder left in the redacted document - proving, field by field, that the redacted layer derives from the same sealed source without needing to trust the party that produced the redaction.

Executive Summary

JIL Sovereign's court-ready evidence bundle generator produces two layers from the same source records: a sealed, unredacted layer preserving full fidelity for restricted-access legal custody, and a redacted, court-facing layer for broader distribution and discovery. In the redacted layer, each field that contains protected information is replaced in place - at the exact position it occupied in the document structure - with a placeholder comprising a truncated cryptographic hash of that field's original value, such as [REDACTED-PHI sha256:<16-hex-chars>].

Because the placeholder is a hash of the specific value it replaces, not a generic blackout mark, any single field's redaction can later be verified: given the disclosed original value, a verifier recomputes the hash and confirms it matches the placeholder left in the redacted document, proving field by field that the redacted layer is a faithful, unaltered derivative of the sealed original.

Core Innovation: Field-level, structure-preserving redaction where each placeholder is a cryptographic commitment to the exact value it replaces - enabling independent, field-by-field proof that a redacted document faithfully derives from its sealed original without disclosing any protected value.

Problem Statement

Conventional document redaction - whether manual blackout, PDF redaction tools, or e-discovery platform redaction logs - operates at the document or block level. The redacted output carries no cryptographic relationship to the original beyond a procedural assertion that "this was redacted from that," which a court, opposing counsel, or auditor must simply trust. There is no way, without separate access to the original unredacted document, to verify that a specific redacted field genuinely corresponds to a specific original value, or to confirm that the redaction process did not also silently alter unrelated, non-protected content elsewhere in the document.

16-hex

Truncated hash length in each placeholder

Layers generated from the same source (sealed + redacted)

1:1

Field-to-placeholder structural correspondence

Why Existing Approaches Are Insufficient

Manual / visual blackout redaction: No cryptographic trace connects the redacted mark to any specific original value; correctness relies entirely on the redactor's diligence.
PDF redaction tools (e.g., generic Acrobat-style redaction): Operate at the visual/block level; provide no field-level, independently verifiable proof of fidelity to an original.
E-discovery platform redaction logs (e.g., Relativity/Concordance-style workflows): Track who redacted what procedurally, but the log itself is not a cryptographic proof binding the redacted output to the specific original field value.

Technical Architecture

Field Type	Sealed Layer Treatment	Redacted Layer Treatment
Protected (PHI/PII-bearing)	Full original value retained	Replaced in place with `[REDACTED-PHI sha256:<16-hex>]`
Non-protected	Full original value retained	Full original value retained, unchanged

Placeholder Construction

For each protected field, the system computes a cryptographic hash of the field's original value and truncates it to a fixed-length hexadecimal representation, embedding it into a structured placeholder string that occupies the exact position the original field held in the document.

Verification Function

Given a disclosed original field value and its corresponding placeholder, a verifier recomputes the truncated hash of the disclosed value and confirms it matches the hash embedded in the placeholder - proving the redacted layer is a faithful, unaltered derivative of the sealed original at the level of that individual field, without revealing the value of any other protected field.

Dual-Layer Bundle Generation Flow

Step	Action
1	Ingest source records for the evidence bundle.
2	Generate the sealed, unredacted layer at full fidelity.
3	Generate the redacted layer from the same source records, substituting each protected field in place with its hash-commitment placeholder.
4	Bind both layers together via a shared bundle hash, so the redacted layer can be proven to derive from the same sealed source as the unredacted layer, rather than a separately fabricated document.

Prior Art Differentiation

Approach	Granularity	Cryptographically Verifiable?	Structure Preserved?
Manual / visual blackout	Block or page	No	No
PDF redaction tools	Block	No	Partial
E-discovery redaction logs	Document, procedural	No (log-based, not cryptographic)	Yes
JIL Sovereign Per-Field Redaction Commitments	Individual field	Yes, per field	Yes, in place

Patent Claim

Independent Claim 44: A computer-implemented method for generating a verifiably faithful redacted derivative of a sealed document containing protected fields, comprising: generating a sealed unredacted evidence layer from a set of source records, at least a subset of fields in the unredacted layer containing protected information; generating a redacted evidence layer from the same source records by, for each field containing protected information, computing a cryptographic hash of that field's original value and replacing the field's value, in place within the document structure, with a placeholder comprising a truncated representation of the computed hash, such that the redacted layer preserves the structural position of every redacted field; and providing a verification function that, given a disclosed original field value and its corresponding placeholder, recomputes the truncated hash and confirms a match, thereby proving the redacted layer is a faithful, unaltered derivative of the sealed unredacted layer at the level of an individual field without revealing the values of any other protected field.