BRENNEN

Industries / Finance and business services

Records management

Records management is the work of keeping an organization's records, from paper files to email and chat messages, so that they can be found when needed, are kept as long as the law or the business requires, and are then destroyed on schedule, unless a lawsuit or investigation requires them to be held.

1950

The US Congress passed the Federal Records Act of 1950. By the National Archives' account, it gave the General Services Administration authority over records' "creation, storage, use, and disposition", let it inspect agencies' records management programs, and had it create records centers for records "that must be preserved for a period of time". The act followed a 1948 report on federal records that Emmett J. Leahy wrote for the Hoover Commission.

prologue.blogs.archives.gov

31

Milestones achieved across 6 domains.

9

To watch for: announced, not yet confirmed outside.

12

Open deliverables. Nobody has shipped them.

In one minute

Where it started

Offices have always kept files, but records management became a named job in the United States after the Second World War. Emmett Leahy's 1948 report for the Hoover Commission gave one estimate that the federal government held 18.5 million cubic feet of records and spent more than $1.2 billion a year making and keeping them. The Federal Records Act of 1950 followed. In 1951 Herman Knaust opened Iron Mountain Atomic Storage in an old iron mine in New York state, and commercial records storage grew from there.

Where it stands

Paper is giving way to digital records. Since 30 June 2024 the US National Archives and Records Administration (NARA) has accepted new transfers of paper records only by exception, and a 2023 rule lets agencies destroy paper originals once they have digitized them to set standards. The hardest cases now are messages and AI tools. In 2022 the US Securities and Exchange Commission (SEC) charged 16 firms over employees' business texts on personal phones, with penalties of more than $1.1 billion. In 2025 a US court made OpenAI preserve ChatGPT logs for a copyright lawsuit.

What is open

Most of the open jobs are tests that nobody has run. Vendors sell AI that sorts records into retention categories, but their accuracy figures come from their own benchmarks. There is also no public test of whether deleted records really leave cloud services, or whether a record destroyed on schedule also leaves the AI tools that have indexed and summarized it. Nor is there one of whether legal holds stop every route to deletion. Each of these tests can be built from synthetic records, so no real person's files are involved.

Domains

AchievedWatch forOpen

Narrower industries

Document shredding

Advances to build with

2 research notes

Open in records management

3 of 12 open deliverables

Open · Storage

A test for locked storage

An open-source test suite that tries to break the retention locks on object storage: Amazon S3 Object Lock, Azure immutable blob storage, Google Cloud Storage bucket lock, and the open-source MinIO and Ceph. For each one it tries every route a user, an administrator or a script has to delete or change a locked test record, such as deleting or overwriting the file, shortening the lock, lifting a legal hold, changing lifecycle rules, deleting the bucket and closing the account, and it records which attempts succeed.

Built with what exists

low-cost accounts on each cloud; MinIO and Ceph run locally in Docker; each provider's own software development kit; pytest to run and report the attempts; Browser Use 1.0, an open-source agent that writes Python to drive Chrome, for routes that exist only in web consoles (see the Note "Deleting 80% to start over").

Credit · unclaimed

Open · Finding records

Find every record, measured

A public test of whether AI search tools find every record that matches a request, as a records officer must for a freedom-of-information request or an audit, not just enough records to answer a question. Generate a synthetic organization's email, chat and documents with planted sets of matching records, write 100 requests, and judge every document against every request so that recall can be counted exactly.

Built with what exists

EnterpriseRAG-Bench's open generation framework, to make the corpus; the search in Microsoft 365 Copilot and in open-source tools such as Elasticsearch; AI agents built on models such as Claude Fable 5.2 (see the Note "Three working devices"); Blair's sampling method for estimating how many relevant documents a search missed.

Credit · unclaimed

Open · Scanning

A checker for digitized pages

Software that checks every page image from a digitization project against NARA's 2023 standards before the paper is destroyed. It looks for low resolution, cut-off edges, skew, blank or missing pages, pages out of order and unreadable text, and writes a report that a second person can audit.

Built with what exists

OpenCV, an open-source image library, for measurements; the Federal Agencies Digital Guidelines Initiative (FADGI) image-quality guidelines on which the rule draws; a model that reads images to compare each page with its neighbors and spot gaps; boxes of printed test documents.

Credit · unclaimed