What should have been an ordinary arbitration filing has become another artificial intelligence embarrassment for the Department of Health and Human Services. The department submitted an argument alleging mistakes by an arbitrator, a routine type of legal document that normally would attract little public attention.

Instead, the filing drew scrutiny from the Federal Labor Relations Authority because it contained a remarkable collection of citation problems. The federal labor agency found enough errors to suggest that artificial intelligence had been used to draft the department’s brief.

The Washington Sun reported that the authority identified 10 “entirely fictitious” citations. Of the filing’s 29 footnotes, 22 were deemed problematic, while 35 of its 42 legal or record citations were found to be “fictitious, erroneous, or irrelevant.”

The problems did not stop with nonexistent or irrelevant citations. The filing also included incorrect page numbers, case numbers, and other details, producing the sort of bureaucratic face plant that basic review presumably was supposed to prevent.

The episode fits a broader record of Trump administration officials embracing artificial intelligence without consistently verifying the results. In July, a State Department representative delivered a presentation at an AIDS 2026 conference that included a map of Africa with multiple countries mislabeled.

Reuters reported that the map contained an artificial intelligence watermark indicating that it had been created with OpenAI tools. An image meant for an international conference instead became another example of an administration presentation undermined by its own questionable material.

Other administration officials have experienced similar problems. Border Patrol officials were caught misusing ChatGPT last fall, while Tulsi Gabbard, then the director of national intelligence, used an AI program to decide which documents related to the 1963 assassination of President John F. Kennedy should be released.

That process did not go well, according to the source. Yet Secretary Robert F. Kennedy Jr.’s Department of Health and Human Services has accumulated more than its share of AI related humiliations, with the latest arbitration filing adding to an already awkward list.

Last year, the department produced “The MAHA Report: Making Our Children Healthy Again.” That report cited multiple scientific studies that did not exist after staff members misused artificial intelligence, leaving a major department document supported by fabricated research.

Soon afterward, the Food and Drug Administration unveiled an artificial intelligence tool intended to accelerate drug approvals. The tool did not work, turning another promised use of the technology into yet another public embarrassment for Kennedy’s department.

The department stumbled again as recently as last month, when it used nonexistent citations generated by artificial intelligence to support sweeping changes to federal teen pregnancy prevention programs. That error surfaced in federal court rather than through the department’s own review.

The Washington Post reported that a federal judge found seven cited articles in the department’s materials. Two were entirely invented, while three others did not exist in the journals to which HHS had attributed them.

The Federal Labor Relations Authority offered the department a direct warning about using the technology in legal proceedings. “While nothing prohibits parties from using AI tools when preparing filings for submission to the Authority, we caution those who do so to be aware that AI continues to present significant risks of error,” the authority said in its ruling.

That warning leaves room for parties to use artificial intelligence while placing responsibility on them to verify what the technology produces. In this case, the authority’s findings show how extensively the department’s filing failed that test.

The filing now joins the MAHA report, the Food and Drug Administration tool, and the teen pregnancy prevention citations as another apparent AI failure at HHS. The department has received a clear reminder about the risks, though its recent history offers little reason to expect the lesson will finally stick.