Government SI Register · STATE · High impact

AI-Augmented Declassification Review

The Department of State lists "AI-Augmented Declassification Review" in the 2025 federal inventory as a high impact SI use case at the deployed stage, in A/SKS. In the agency's words, the problem it is meant to solve: "The amount of documents, particularly cables and emails, that require declassification review increases exponentially in the next few years. Manual review is unsustainable and expensive given the number of cables (in the hundreds of thousands) and emails (increasing from hundreds of thousands to millions)." Every field below is quoted as filed.

ID none givenStage DeployedSource row 2835

The record, as filed

Every field the agency reported

Bureau or component
A/SKS
Stage
Deployed
High impact status
High-impact
Topic area
Admin Functions
Technology type
Classical ML
Problem it is meant to solve
The amount of documents, particularly cables and emails, that require declassification review increases exponentially in the next few years. Manual review is unsustainable and expensive given the number of cables (in the hundreds of thousands) and emails (increasing from hundreds of thousands to millions).
Expected benefits
The expected benefits and positive outcomes from using AI are cost savings by reducing the need of manual reviews, reduce labor in cable review by up to 80%, reduce the time needed for annual review, and create more consistency in the review process.
What the system outputs
The AI system's outputs are binary classification predictions for documents on whether a document should be declassified or exempt from declassification and multiclassification for reasons for exemption.
Operational or pilot start
02/02/2023 00:00:00
How it was built or bought
Contracting and In House
Vendors
Deloitte
Authorization to Operate
No
Data used to train, tune or evaluate
The data used to train the model are cables from 1995-1999 that have completed manual review with metadata on decisions from manual declassification review. Additional data includes classification/declassification guides and associated glossaries to improve model performance. Performance evaluation is measured by a human Quality Control reviewer.
Uses personally identifiable information
No
Demographic features used
k) None of the above
Custom-developed code
Yes
Pre-deployment testing
Yes
Impact assessment completed
In-progress
Potential impacts identified
The model could incorrectly predict to declassify a document. The model could predict to exempt a document that should have been declassified, reducing public visibility. All exempted documents are reviewed by a human.
Independent review
In-progress
Ongoing monitoring
Yes - Monitoring Established
Operator training
Training Established
Fail-safe
Yes
Appeal process
Appeal Process Established
Consultation with users and the public
Usability Testing

Left blank in the record: Justification given, System name, Federal Data Catalog link, Privacy Impact Assessment, Public source code. Point of contact details are in the source file.

Source

Source: row 2835 of 2025_individually_reported_AI_use_cases.csv in OMB's 2025 Federal Agency AI Use Case Inventory repository (summary as of April 13, 2026), read Nov 17, 2026. File SHA-256 a62721202ca4b110

Life areas: Government operations.

Records, not ratings. Every entry is what an agency reported to OMB, quoted as filed. The register does not test, rate or vouch for any system, and a record does not prove how a system is used day to day.