VA higher-level reviewers sent back 35% of decisions in fiscal 2026
Error counts and accuracy measures
The U.S. Department of Veterans Affairs counted more than 329,000 errors collecting medical evidence needed to process veterans’ disability claims in fiscal year 2026, compared with more than 66,000 in fiscal year 2021. The VA introduced Automated Decision Support in December 2021, and the evidence-error tally climbed in each subsequent fiscal year. The Washington Bureau of The Beam, an accountability newsroom at Arizona State University’s Howard Center for Investigative Journalism, found the agency’s growing use of the system coincided with a surge in errors made while collecting medical records. The VA denies that automation contributes to errors. The department runs one of the federal government’s largest disability programs, providing treatment and compensation for millions of veterans for injuries or conditions connected to their service.
The VA’s higher-level reviewers sent back about 23% of claim decisions to collect more evidence and reconsider them in fiscal year 2021, just before Automated Decision Support launched. Five years later, the rate reached 35% in fiscal year 2026. Reviewers may return a decision because they disagree with the original determination or because of evidence-collection errors. Those errors can stem from automated or human mistakes, and the statistics the VA publishes do not distinguish between the two. Either way, the VA is mandated by law to help veterans prepare documentation needed to substantiate their claims.
Margarita Devlin, the VA’s principal deputy under secretary for benefits, testified before the House Committee on Veterans’ Affairs that technology had helped streamline claims processing, decision accuracy had improved and errors had decreased. She used an issue-based accuracy metric that scores correctness on each condition in a sample of claims containing both denials and approvals. Devlin said the 12-month accuracy rate was close to 94% at the time of her April 15, 2026 testimony, below the VA’s 96% target and the highest it had been in two years. The VA did not make Devlin available for an interview after repeated requests or respond to questions about her testimony.
Devlin’s metric covered correctness on sampled conditions, while the department’s appeal data separately tracked evidence-collection errors and decisions returned for reconsideration. A duty-to-assist error occurs when a claim proceeds to review without a necessary examination, a lapse in the VA’s legal mandate to help veterans obtain evidence supporting their claims. Such errors are tracked only when a veteran appeals a claim decision. Many veterans do not appeal, so the scale of the problem could be much higher than the reported error count indicates.
The role and expansion of Automated Decision Support
Automated Decision Support does not determine how much compensation a veteran receives and is used only for certain medical conditions. When a veteran submits a claim for a condition approved for automated processing, the tool collects medical records from the VA and records from the applicant’s time in service, scans the files for keywords and data points, and writes a summary. Before the file reaches a human benefits rater, Automated Decision Support analyzes the records and decides whether the claim is ready for review. If it finds insufficient evidence, the tool orders an additional examination to help the veteran support the claim. Human raters use the generated summary as a reference when deciding the severity of an injury, how much it stems from military service, and the monthly compensation owed.
The VA expanded Automated Decision Support to regional offices across the country and to additional conditions. Budget documents show that the tool made nearly 300,000 decisions on claims in 2023, and by April 2026 its pace had reached as high as 50,000 per month. The VA’s latest annual budget request calls for applying the tool to 90% of all claims by the end of 2027. The department says the system speeds processing as it confronts long-standing wait times and backlogs of unprocessed claims.
Evidence from claims, reviews and appeal decisions
The Beam reported that Randy Leger Diaz’s experience had become more common as the VA increasingly used automation to gather veterans’ records and summarize medical evidence before claims determinations. Leger, a former Marine, said his whole body was in pain when he left the service. Following rigorous physical training in the mornings, he spent his days in uniform working in a warehouse in Okinawa, Japan, where he was part of the 7th Communication Battalion from 2014 to 2016. After returning to civilian life, Leger said, his joints would pop and lock, his back, wrist and shoulders ached, sharp pains shot up and down his legs, he couldn’t sleep and he had stomach problems. “They always tell us in the Marines to suck it up,” he said. That advice worked for him in his 20s, but in his 30s, he said, his injuries started to affect his ability to work.
Leger first filed for VA benefits in November 2024, seven years after finishing his service at Parris Island in South Carolina. The VA denied him monthly compensation for most of his conditions. Leger said his attorney requested a higher-level review. After the review, the VA told him it had committed errors collecting medical evidence and failed its legal duty to assist veterans in preparing their claims. Leger said he eventually received monthly compensation for some of his rejected claims.
Leger posted screenshots on Reddit showing VA notices identifying duty-to-assist mistakes on his claims. On the VAClaims subreddit on Jan. 7, 2026, he posted: “Basically every single issue they previously denied or mishandled now magically has an error attached to it.” He added, “I don’t even know whether to laugh or be angry anymore.”
The American Legion conducted reviews in 2025 at VA regional offices in Chicago, Puerto Rico and Louisville, Kentucky. Employees at those offices reported that the automated tool frequently missed details and failed to extract critical evidence from documents, according to the reports. The San Juan report said the “limitation frequently results in partial rather than complete claims being submitted” and “delays and contributes to the underdevelopment of claims.”
The Board of Veterans’ Appeals publishes documents explaining its decisions, including two cases cited by The Beam. In a 2024 case, the VA proposed reducing a veteran’s disability compensation for bilateral hearing loss; a review found an automated system had failed to identify relevant evidence within the veteran’s record, and the disability rating increased after additional medical records were submitted. In a 2025 case, the board found “significant discrepancies” between an automated summary and the veteran’s actual service treatment records. The VA had denied a service-related asthma claim without providing an asthma examination, and an appeals judge found a duty-to-assist error and sent the claim back for reconsideration.
Competing explanations for the increase
VA press secretary Quinn Slaven said, “There is no evidence to suggest the Automated Decision Support tool is causing errors.” He attributed the rise in identified errors to the number of denials appealed through the higher-level review process quadrupling from 2020 to 2026. Slaven said the quadrupling “is why more errors are being found at this stage of the process.” The Beam presented Slaven with data showing that the share of higher-level reviews sending cases back to be reconsidered rose alongside the total number of errors. Slaven did not address those statistics in a response to detailed questions.
Devlin testified that technology had streamlined claims processing, improved decision accuracy and reduced errors. “Our errors, in general, are decreasing and our percentage of accuracy has been going up,” she testified April 15, 2026, under questioning by the House committee’s senior Democrat, Rep. Mark Takano of California. Takano told The Beam by email: “A quick decision is well and good. But if we’re moving so quickly that errors are made, or if we’re simply rushing to denials, veterans are not being served well.”
According to The Beam’s investigation, employees have told Congress and veterans advocacy groups that Automated Decision Support routinely fails to extract key details from documents and misunderstands medical nuances. A Democratic aide on the House Veterans’ Affairs Committee, speaking by email, acknowledged that appeals had increased but said the “number of errors found in Higher Level Reviews or caught by the Board of Veterans’ Appeals are also increasing.” The aide said, “That speaks to a quality deficit” at the Veterans Benefits Administration, the VA subagency that handles benefits, and “That can be directly attributed to the errors produced by ADS.” Attorney John Niles, who specializes in veterans appeals, said by email, “There can be no doubt that VA’s expanding reliance on Automated Decision Support is a contributing cause of duty-to-assist errors.” The VA disputes the connection.
The story was originally published by The Beam and distributed through a partnership with The Associated Press. The Beam is an accountability newsroom inside the Howard Center for Investigative Journalism at Arizona State University.