1.5 million FOIA requests in FY2024. A 267,000-case backlog growing 33% year over year. Complex requests can take dozens or even hundreds of staff-hours by hand. Re-Doc automates the detection step. Officers review and approve instead of starting from a blank page on every document.
Incident type: Public disturbance. No injuries.
FOIA officers are processing documents one page at a time. Complex requests can take dozens or even hundreds of staff-hours. Multiply by 1.5 million annual requests. Then add the lawsuits when redactions are done wrong or when visual overlays leave underlying text exposed.
The government-wide FOIA backlog grew 33% in FY2024 to 267,000 unprocessed requests. Federal agencies processed over one million requests but received 1.5 million. The gap is operational, not legal.
Agencies have the authority to release. They have the staff. They do not have enough hours. Manual redaction cannot scale to the volume of incoming requests. The backlog compounds every year while requesters with legitimate rights to public records wait months or years for responses.
Using annotation or drawing tools in PDF software to place black boxes over text only hides the visual information — the underlying text data often remains recoverable via copy-paste. This is the same failure that exposed 22,391 SSNs in PACER. U.S. Courts guidance explicitly warns against these methods. For government records, releasing content where the exempt information is still accessible can constitute an improper disclosure under FOIA and the Privacy Act, opening the agency to litigation.
True redaction requires a tool that structurally removes the exempt content from the document file — not just an overlay that hides it visually. The text data stream must be empty, not just covered.
The average FOIA response involves multiple documents, multiple pages, multiple exemption categories. A FOIA officer reads every page, identifies exempt content, and marks redactions manually. For large, complex requests involving voluminous records, multi-custodian reviews, or frame-by-frame analysis, this can require dozens or even hundreds of staff-hours per bundle.
Automation replaces the detection step. The officer's legal judgment on exemption categories remains human — that cannot be automated. The hours come from the mechanical work of finding every name and address on every page.
Body worn camera release requests are not just video. They include incident reports, arrest records, case notes, and witness statements containing bystander names, addresses, and personal details. Per BJA/OJP guidance, BWC public release packages typically include multiple companion documents — incident reports, officer narratives, and witness statements — each requiring separate review and redaction. Video tools like Axon and Motorola handle the footage. They do not handle the paper trail.
The video gets blurred. The incident report goes out with the officer's name, the bystander's address, and the witness's phone number intact. That is the gap Re-Doc closes.
Re-Doc sits between document collection and release. Upload batches via API. Get back every page with exempt information flagged and covered. Officers review AI detections and approve.
FOIA request received. Responsive records uploaded via batch API or individual file submission. Any format.
Visual processing reads every page, scanned or native. A context-aware model identifies names, SSNs, addresses, informant IDs, law enforcement techniques, and all other exempt categories. Redactions applied.
Officer reviews AI-flagged redactions. Adjusts as needed. Applies exemption codes. Approves the release package. Reviewing is faster than finding from scratch.
Documents released with structural redactions. Underlying text removed from the file, not just overlaid. Cannot be copied, extracted, or reversed. Meets FOIA's 'reasonably segregable' requirement.
Some documents need permanent redaction for public FOIA release. Others need to stay fully readable: for inter-agency sharing, AI training, or internal review. The right pipeline depends on who receives the document.
Structural text removal. Required for documents going to the public.
Visual processing reads the scanned record, finds exempt content by region, and burns permanent black boxes into the output PDF. The underlying text is removed, not just covered. To comply with FOIA's “reasonably segregable” requirement, the released copy must not reveal the exempt information. This pipeline produces output designed to meet that standard.
Synthetic data swap. Document stays fully readable.
Replaces real identifiers with synthetic equivalents that are statistically plausible but not real. The document remains fully readable and structurally intact. No black boxes. Used when the recipient needs to work with the document content, not just receive a redacted copy.
The case number stays. The file structure stays. Every personal identifier is permanently removed — not covered with a box that still holds the text underneath.
Case No: CASE-2024-08812
Subject: Thomas R. Calloway
DOB: June 8, 1990
Address: 3412 Birch Road, Sacramento CA
Informant: CONF-INFRM-001
Officer: Martinez R.
Case No: CASE-2024-08812
Subject: Thomas R. Calloway
DOB: June 8, 1990
Address: 3412 Birch Road Sacramento CA
Informant: CONF-INFRM-001
Officer: Martinez R.
Case number preserved for recordkeeping. Underlying text structurally removed. Meets FOIA's “reasonably segregable” standard.
Different agencies, different request types. Same operational problem. Every responsive document still requires an officer to find what needs to be covered before anything goes out the door.
A federal agency receives 200 documents responsive to a single complex FOIA request. Under current tools, a FOIA officer reads every page to identify exempt content: names, SSNs, informant identities, law enforcement techniques. Then marks redactions manually. With Re-Doc, the 200 documents go through the Batch API. The officer receives pre-flagged documents, reviews the AI detections for accuracy against the applicable exemptions, and approves release. The product is designed to support agency workflows for compliance with 5 U.S.C. § 552. The legal judgment on exemptions remains the officer's responsibility.
State Public Records Acts and local Open Records Laws create the same volume problem at every level of government: city police departments, county agencies, state health departments. A city police department handling hundreds of BWC-related public records requests per year has the same manual bottleneck as a federal agency. Re-Doc processes incident reports, inspection records, permit files, and administrative documents under state open records obligations the same way it processes federal FOIA packages.
A BWC release request triggers both video review and document review. Axon, Motorola, and VIDIZMO handle the video footage. Nobody handles the incident reports, arrest records, and witness statements that typically accompany the release. Per BJA/OJP guidance, these companion documents contain bystander names, witness addresses, medical information, and officer-identifying details that require separate redaction. Re-Doc processes the document side of every BWC package. The video tool handles the footage. The two work side by side.
Most tools in the government FOIA space treat document redaction as a manual, per-page task. Re-Doc automates the detection step at scale.
Typical tools
in the market
Re-Doc
Purpose-built for government
Improper redaction methods leave underlying text recoverable via copy-paste.
Structural text removal. The exempt content is permanently removed from the file, not just overlaid.
Manual, one page at a time. Complex requests can take dozens or hundreds of staff-hours.
Batch API processes entire FOIA bundles in one submission. Officer reviews, not detects.
A significant portion of government records, especially legacy files, are scanned images requiring specialized processing.
Visual processing handles scanned records of any age, quality, or format.
No API access. Every document must be processed manually one at a time.
REST API and batch upload slot into any existing agency document workflow.
Cloud-only deployment. FedRAMP procurement barrier for many federal and law enforcement agencies.
On-premise deployment available. Contact us for agency deployment requirements.
taxpayer cost to process FOIA requests in FY2024
That cost is almost entirely staff-hours. Officers reading documents and drawing boxes by hand. Automation replaces the mechanical detection step. The officer's legal judgment on exemptions remains human. That cannot be automated.
Common exempt categories under 5 U.S.C. § 552, automatically identified across every page for FOIA officer review. Detected in both native text and scanned records via visual processing.
Custom entity types configurable via API. Add agency-specific identifiers, program codes, or classification markers for any record type.
References & Sources