Traci Tamiko Eto, Mayo Clinic’s former director of research operations, filed a federal whistleblower lawsuit on July 6, 2026 in U.S. District Court in Minnesota alleging that staff working on the hospital system’s AI-integrated digital assistant, known as MAYA, knew the tool had an error rate as high as 67% and worked to keep that number from surfacing, according to reporting by MPR News and Futurism. The suit alleges unapproved software was pushed into live clinical workflows and that Eto was pushed out of her job after repeatedly raising concerns internally.
What the lawsuit alleges MAYA got wrong
According to the filing, MAYA was built as an AI-integrated assistant intended to support clinical and administrative workflows at Mayo, one of the country’s most prestigious academic medical systems. Eto’s complaint alleges that internal testing turned up an error rate as high as 67% for the tool, and that rather than disclosing or fixing that failure rate, staffers on the project deleted unflattering test results and mischaracterized what the tool could reliably do. The complaint reportedly cites ten additional whistleblower reports from other Mayo staff raising the same core allegation: that people working on the underlying study were trying to suppress the error rate from becoming known internally or externally.
What happened to the person who raised the alarm
Eto’s suit states she filed ten separate whistleblower complaints about MAYA’s problems before facing retaliation that culminated in her departure in early 2025. According to coverage of the filing, she alleges she was excluded from executive meetings, told she was a “poor cultural fit,” and ultimately given a choice between resigning or having her personnel file altered in ways that would have made her unemployable elsewhere in health care. Whistleblower retaliation claims of this kind, if substantiated, typically fall under federal protections designed to shield employees who report patient-safety or fraud concerns from adverse employment action.
Why this case matters beyond one hospital
Legal analysts describe this as one of the first major federal whistleblower cases centered specifically on AI governance inside a health system, rather than on a vendor’s product. That distinction matters: Mayo built and deployed MAYA itself rather than licensing it from an outside AI company, meaning the allegations point directly at how a major health system managed its own internal AI development and testing pipeline, not at a third-party vendor’s marketing claims. If the allegations hold up, they would suggest that even a system with Mayo’s institutional resources and clinical reputation can face the same pressure to rush an under-validated AI tool into use that smaller, less-resourced hospitals face with vendor products.
Mayo’s response and the broader pattern of AI safety disputes
Mayo Clinic has publicly disputed the characterization of MAYA’s performance and its handling of the internal concerns, according to coverage of the litigation, though the health system’s substantive rebuttal to the specific 67% figure has not been independently verified through peer-reviewed data. The dispute echoes a broader pattern seen across health-system AI deployments in 2026: internal pilot data on AI accuracy is rarely published in peer-reviewed journals, leaving patients, regulators and even other clinicians within the same institution dependent on internal assurances rather than independently auditable evidence.
The regulatory gap this case exposes
Unlike an FDA-cleared diagnostic algorithm, an internally built assistant tool like MAYA used for workflow support may fall outside formal medical device regulation depending on its specific function, leaving oversight largely to a hospital’s own internal governance — the exact process this lawsuit alleges failed. That gap is part of why frameworks like URAC’s new Artificial Intelligence in Health Care Accreditation and the Coalition for Health AI’s (CHAI) assurance standards have gained traction in 2026, as health systems look for external validation that internal AI governance is not, as this lawsuit alleges happened at Mayo, quietly overridden by internal pressure to ship.
What’s next
The case remains in early stages in federal court in Minnesota, and Mayo has not disclosed whether MAYA remains in active clinical use following the allegations. Discovery in the case could surface the internal testing data at the center of the dispute, which would give outside researchers their first real look at whether the 67% error rate allegation is accurate and, if so, what specific clinical or administrative functions MAYA was performing when it failed. Health law observers expect the outcome to influence how other health systems document, and defend, their internal AI testing going forward.
