WASHINGTON, D.C. (Government Executive) — Faced with a backlog of accommodation requests from employees with disabilities and medical conditions, the Labor Department plans to use artificial intelligence to triage the workload.
An internal email obtained by Government Executive states that staff at the Civil Rights Center, which processes reasonable accommodation requests, currently must “engage substantively with every request regardless of completeness or likely outcome, limiting capacity and delaying determinations.”
“The process requires case-by-case judgment, but cases vary significantly in complexity and documentation quality: some are well-supported and straightforward, some require additional documentation, and some are not well-supported by the documentation provided,” the email states.
Labor Department employees told Government Executive that the department is facing a backlog of hundreds of RA requests. Several reported waiting months or, in some cases, a year or longer for for reviews, delays they say have hampered productivity at a department charged with protecting the workplace rights of workers with disabilities.
One DOL employee told Government Executive that he submitted a reasonable accommodation request in March 2025, supported by documentation from his primary care and a mental health providers, yet his case remains pending.
“It’s just kind of ironic to me that the Labor Department is violating policies that they set,” he said. “That’s the most frustrating part, to be honest, is that they can’t even abide by their own rules.”
Labor officials wrote in the internal memo that AI will “streamline” this work, citing four AI applications under review “to improve this process.”
The memo stresses these AI tools will play a supporting role in processing reasonable accommodations, and that Civil Rights Center employees “will continue to make decisions on necessary actions (including how and what is requested).”
The department expects that AI “could triage incoming requests by likely complexity and documentation sufficiency, allowing staff to prioritize effectively.” It adds that “AI-assisted document analysis” could flag incomplete or inconsistent medical documentation earlier in the process, “reducing back-and-forth and shortening timelines.”
The memo also states that “Al-enabled communication capability could engage requestors by providing updates, prompting requestors for additional information, and initiate follow-up to keep requests progressing without staff intervention.”
DOL also wrote that “Al could assess whether the accommodation requested is effective for the documented functional limitation,” or suggest “if other equally effective options exist.”
The Labor Department did not respond to a request for comment.
Jodi Hershey, a former reasonable accommodation specialist at the Federal Emergency Management Agency and the founder of the workplace accommodation firm EASE, LLC, said “the problem DOL is trying to solve is real” – many federal agencies don’t have the staff needed to keep up with the volume of incoming accommodation requests.
“I’ve seen employees wait a year or more for a decision on something they needed on day one — sometimes going without the very accommodation that lets them do their job. So I don’t dismiss the impulse to find a faster path. The status quo is already failing disabled employees,” she said.
Hershey said AI “could genuinely help employees” navigate the reasonable accommodation process. “But that’s very different from an agency using AI to take the human judgment out of the decision, and that’s where this plan worries me,” she said.
“‘A human still makes the final call’ sounds reassuring, but if the AI has already sorted the case, flagged the documents, and suggested an answer, the human is grading the algorithm’s homework. Real oversight means a person can actually reach a different conclusion — and that the employee knows a machine was ever involved,” she said.
Michael Fallings, managing partner at Tully Rinckey PLLC, a firm specializing in federal employment law, said he’s represented clients who have waited a year or longer for their agencies to process their reasonable accommodation requests.
While agencies may be tempted to use AI to speed up the work, Fallings warned that processing confidential medical records through AI tools may raise some legal risks.
“I’m sure the government’s thought process is … ‘We may not have the number of staff to handle it,’ but that does then create privacy issues for employees when you’re not using people, you’re using technology,” Fallings said.
Hershey said she was most concerned by the Labor Department’s plans to triage requests by “likely outcome” using AI.
“The employees with the thinnest documentation are often the ones with the least access to good healthcare, or to anyone who can help them navigate the paperwork. If the system quietly moves those cases to the back of the line, you’ve built a process that works best for the people who needed the least help to begin with,” she said.
