Rising FOIA Workloads Drive Interest in AI With Human Oversight

Federal FOIA programs are facing rising workloads alongside staffing constraints, technology changes, and other resource limitations. In FY2025, U.S. federal agencies received about 1.7 million requests, a 13% increase from FY2024 and the highest total on record. The U.S. Department of Justice (DOJ) received 159,743 requests—more than 27,000 above its FY2024 total—and its backlog increased from 21,567 to 29,309. These figures create a practical case for agencies to assess whether existing intake, processing, and disclosure workflows can handle both current demand and further growth.

The operational pressure is significant because FOIA generally requires agencies to make a determination within 20 working days. One analysis cited in the source found that requesters with pending submissions waited, on average, more than 240 days to hear back. Delays can increase litigation risk, adding cost and workload for already stretched teams. For procurement professionals, the reported volumes and backlog are useful planning context when evaluating workflow capacity, case-management requirements, and support for records review. The source does not identify a specific solicitation, award, or funding opportunity, so these pressures should be treated as modernization signals rather than an announced procurement.

Artificial intelligence could support parts of the process, including request routing, document summarization, classification, first-pass analysis, and identification of potentially responsive material for further review. The recommended approach is to use AI for bounded tasks with known source material and verifiable outputs, while preserving clear human accountability. As a tool’s role moves closer to consequential determinations, the need for human oversight increases. Agencies and prospective contractors should therefore be prepared to explain how outputs are reviewed, how task boundaries are enforced, and how decisions can be audited rather than presenting automation as a substitute for accountable FOIA professionals.

AI may also affect demand from the requester side by making it easier to research agencies, draft requests, and submit them at scale. The source describes an anecdotal report of 50–100 requests arriving within moments from one requester, alongside similar accounts from other practitioners; it does not establish this as a systemwide trend. Still, the possibility reinforces the need to plan for intake capacity as well as processing efficiency. FOIA modernization designs should incorporate legal, security, privacy, and records-management controls from the outset. The source identifies Casepoint as a relevant provider whose offerings may align with these priorities, but it does not name Casepoint in connection with a specific government award or procurement.