67 Percent, and the Institutional Gate That Wasn't Checked

The exhibit, as reported by Futurism citing Minnesota Public Radio: tucked inside ten separate whistleblower complaints, the allegation that the team working on MAYA — Mayo Clinic's AI-integrated digital assistant — knew the tool had an error rate as high as 67 percent.

The pattern is the rest.

A clinical AI tool does not keep operating with a 67 percent error rate because engineering is broken. It keeps operating because the people meant to catch it are not in the room. The named federal gate is the Institutional Review Board — the body that weighs whether a device touching patients is safe to test and later safe to deploy. Per the complaint, Mayo allegedly authorized a cardiac surgical procedure using MAYA without adequate IRB review. If the IRB was bypassed, the question is not whether the model works. It is who let it ship without the box being checked.

Per the same Futurism report on MPR, the former director of research operations who tried to enforce that review — Traci Tamiko Eto — was placed on a performance plan in spring 2025, told she was a "poor cultural fit," and offered a choice: resign, or face personnel-file alterations "that would render her unemployable at Mayo and would impede her career outside the institution." Her supervisor's reported response to her privacy concerns about Mayo Clinic Platform: fixing them would "jeopardize the pace on ongoing research projects, which in turn would compromise Mayo's competitive advantage."

Mayo's public posture, per the same reporting: "committed to responsible development and deployment of AI." The complaint was filed July 6, 2026 in U.S. District Court for the District of Minnesota. The portable question: who is the named safety reviewer, and what is the retaliation record for the people who raise flags?