# What Is Redaction in Data Production? | Kodex

Source: https://www.kodexglobal.com/glossary/redaction

> Redaction removes information from records before production. Learn what must be redacted and why visual masking is not redaction.

Request Operations

## Redaction

Redaction is the permanent removal of information from records before they are produced, typically to withhold data about third parties or material outside a request’s scope. Effective redaction removes the underlying data rather than obscuring it visually, and failures here are usually technical rather than editorial.

Redaction is where careful scoping gets implemented, and where it most often fails for reasons that have nothing to do with judgement.

### What gets redacted

- Third-party data. Records about the named account frequently contain information about other people: the other side of a conversation, other names on a shared payment method, other users in a group. Those people are not the subject of the request.

- Out-of-scope material. Records outside the date range or data type specified.

- Material beyond the instrument’s authority. Content appearing in records responsive to a subpoena, for example.

- Privileged material, where applicable.

### The technical failure mode

The recurring problem is that redaction is treated as a visual operation. Covering text with a black rectangle in a PDF, changing the font colour to match the background, or cropping a screenshot in a viewer all leave the original data present in the file.

Visual maskingTrue redaction

Data present in the delivered file Yes No

Recoverable by copy and paste Often No

Recoverable by text extraction Yes No

Survives format conversion No; often exposed Yes

Suitable for production No Yes

The correct approach is to remove the data and then generate the output file from the redacted source, or use tooling that genuinely deletes content rather than layering over it. Verifying by extracting text from the finished file (not by looking at it) is the check that catches the failure.

### Documenting redactions

A production that silently omits material is difficult to distinguish from an incomplete one. Marking that a redaction was applied, and the basis for it, serves the recipient and the producer both: the agency can challenge a redaction it disputes, and the company has a record of a decision it may need to explain much later.

### Why this is a user protection

Third-party redaction is the clearest case. A request concerns one account, but the data held about that account inevitably describes other people who are not suspects, are not named, and have no notice and no opportunity to object. Nobody is advocating for them in the process. Redaction is the mechanism by which their data stays out of an investigation that was never about them, and it only happens if the producing company does it deliberately.

FAQ

### Frequently asked questions

Why is drawing a black box over text not redaction?

Because in most document formats the text remains in the file underneath the graphic. It can be recovered by copying the text, extracting the document’s contents programmatically, or simply moving the overlay. Genuine redaction deletes the underlying data and produces a file that never contained it.

What typically needs redacting from a production?

Information about third parties who are not the subject of the request, records outside the specified date range or data type, material protected by privilege, and content the instrument does not have authority to compel. The common thread is that it was never within scope.

Does redaction risk non-compliance?

Redacting non-responsive material is not withholding responsive material. Where the basis for a redaction might be contested, note that a redaction was applied and the reason, so the agency can raise it rather than discovering a gap later.

Glossary

### Related terms

#### Responsive Records

Responsive records are the documents and data that actually fall within the scope of a specific request. Determining what is responsive is a scoping judgement, and getting it wrong in either direction means either failing to comply or disclosing more of a user’s data than the law required.

#### Chain of Custody

Chain of custody is the documented record of who handled a piece of evidence, when, and what they did with it, from collection through to presentation. An unbroken chain supports the claim that evidence is what it purports to be and has not been altered; gaps invite challenges to its admissibility.

See how Kodex structures review and production.
Learn more