Labor Department explores AI to triage disability accommodation requests amid backlog

The Labor Department is exploring artificial intelligence to address a backlog of hundreds of reasonable accommodation (RA) requests from employees with disabilities and medical conditions. An internal email obtained by Government Executive describes a plan to use AI to triage incoming requests by likely complexity and documentation sufficiency, allowing staff to prioritize cases more effectively.

The email, attributed to Labor Department staff, states that current processes require case-by-case judgment regardless of completeness or likely outcome, limiting capacity and delaying determinations. It notes that cases vary significantly in complexity and documentation quality, with some requiring additional documentation and others not well-supported by the provided materials.

Labor Department employees told Government Executive that the backlog has led to delays of months or, in some cases, a year or longer for reviews. One employee reported submitting a request in March 2025, supported by documentation from primary care and mental health providers, yet the case remains pending. The employee criticized the department for violating policies it sets for workplace rights protections.

Labor officials wrote in the memo that AI tools will play a supporting role in processing RA requests, with employees continuing to make final decisions. The department expects AI to streamline triage by complexity and documentation sufficiency, flag incomplete or inconsistent medical documentation earlier, and use AI-enabled communication to provide updates and prompt requestors for additional information.

The memo also suggests AI could assess whether a requested accommodation is effective for the documented functional limitation or propose equally effective alternatives. However, the Labor Department did not respond to a request for comment on the plan or its implementation timeline.

Experts raised concerns about the potential risks of using AI in this context. Jodi Hershey, a former reasonable accommodation specialist and founder of EASE, LLC, acknowledged the real problem of staffing shortages but warned that using AI to sort cases and suggest answers could undermine human oversight. She emphasized that true oversight requires humans to reach independent conclusions, not merely grade the algorithm's work.

Michael Fallings, managing partner at Tully Rinckey PLLC, highlighted legal risks associated with processing confidential medical records through AI tools, noting that privacy concerns arise when agencies substitute technology for human staff in handling sensitive information.