WASHINGTON — Removing a survivor’s name from a document may not protect their identity. A reference to a relative, a school or a workplace can also identify them. The Justice Department’s instructions for reviewing the Epstein files recognised that risk and required reviewers to look beyond highlighted names.
The January 4, 2026, protocol identified a specific limitation: the review software’s name-highlighting feature did not work in Excel files or images. Elsewhere, it could miss misspellings or highlight names that did not require protection. Reviewers were instructed to read entire documents carefully.
The rules explain what reviewers were supposed to do. They do not show how consistently those instructions were followed—or what happened when reviewers discovered a recurring problem.
The published review protocol directed reviewers to send questions to a central email inbox rather than record them in the document-review platform. If they identified a possible victim absent from the circulated list, they were instructed to send the document number, the person’s name and a brief explanation.
That arrangement is not evidence of concealment. But anyone trying to reconstruct a decision might need both the document’s review history and the related emails.
Those records could help answer practical questions: who flagged a concern, how it was resolved and whether the response led to another check of documents already reviewed.
The department also described additional victim-privacy checks by federal prosecutors in New York and Florida. It acknowledged that inconsistencies were likely in a large manual review and said the additional safeguards were intended to minimise them.
Neither sampling nor inconsistent redactions alone proves negligence. The more useful question is what happened after a problem was found: did the department change instructions for subsequent work, require earlier documents to be checked again, or both?
The protocol also allowed reviewers to classify records as duplicates or outside the release’s scope. Intermediate emails could be treated as duplicates when a more complete version of the exchange had already been reviewed and appropriately redacted.
Those instructions offer a legitimate explanation for some differences between the collected archive and the published files. They do not explain every omission.
An October 28 CBS News investigation identified gaps in document-numbering sequences but could not establish what the absent records contained or why each was missing. CBS reported DOJ’s position that it had released every document required by the transparency law, with remaining material unrelated, privileged or duplicative.
The published rules provide a starting point for examining those decisions. They identify the software limitations reviewers were warned about, the judgments they had to make and where they were supposed to send questions.
The question of review quality is distinct from the Epstein files transparency failures previously covered by The Eastern Herald: one concerns what the government releases, the other how it decides material is safe to release.
No internal escalation logs or batch-approval records were obtained for this analysis. The rules show how the process was designed to work. They do not establish why any particular privacy failure occurred.
