How Mayo Clinic Hid Its AI’s 67% Error Rate: Delete the Test Results

A lawsuit filed in July 2026 alleges that Mayo Clinic deployed an AI medical assistant called MAYA while internal staff knew the tool carried an error rate as high as 67%.

Ranked the number-1 hospital in the country by U.S. News & World Report, Mayo Clinic is facing allegations that it cut corners on AI safety, buried damning test results, and retaliated against the employee who tried to stop it.

The suit was brought by former Mayo research director and AI compliance lead Traci Tamiko Eto, who says she was pushed out after raising alarms. Patients treated at Mayo during this period may have received care influenced by a tool its own developers knew was failing more than half the time.

If you’ve ever trusted a hospital’s reputation the way you’d trust a seatbelt, this one’s going to sit with you.

About the Mayo Clinic AI Platform Fail

According to reporting by Minnesota Public Radio, Eto joined Mayo in 2023 and almost immediately began flagging problems. She raised privacy concerns about the Mayo Clinic Platform, an AI-integrated data system.

Her supervisor’s response, the lawsuit alleges, was to tell her that fixing the issue would “jeopardize the pace of ongoing research projects” and “compromise Mayo’s competitive advantage.”

The research kept moving. The problems stayed unfixed. The MAYA allegations are worse.

The lawsuit claims the team developing MAYA — Mayo’s AI-integrated digital assistant — deleted unflattering test results, misrepresented the tool’s capabilities to leadership, and made data security decisions that put patient information at risk. When internal testing revealed an error rate as high as 67%, the lawsuit alleges, staffers tried to conceal it rather than report it.

To be clear about what 67% means in a medical context: the tool was allegedly wrong more often than it was right. A coin flip would have outperformed it. And the people who knew that, according to the lawsuit, decided the answer was to hide the number rather than fix the tool.

Eto filed 10 separate whistleblower complaints. By early 2025, she says she was being excluded from executive meetings. Then came the ultimatum: Resign or face changes to her personnel file that would, the lawsuit alleges, “render her unemployable at Mayo and would impede her career outside the institution.”

Mayo’s communications director told Minnesota Public Radio the organization “conducts research and clinical innovation in accordance with applicable laws and regulations” and does not comment on active litigation.

That response lands differently when the lawsuit alleges the opposite — that federal review processes for new technology were bypassed, that a tool with a catastrophic error rate was deployed anyway, and that the person who raised the alarm was shown the door.

“In accordance with applicable laws and regulations” is doing a lot of heavy lifting in that statement.

What This Means If You’re a Mayo Patient

The lawsuit centers on Eto’s wrongful termination, not individual patient harm claims. But the implications for patients are real: if MAYA was influencing clinical workflows or patient-facing interactions while carrying a 67% error rate, patients had no way to know a machine was involved — let alone a broken one.

Eto’s attorney, Artur Davis, put it plainly to Minnesota Public Radio:

“If [people] care about the notion that AI has to be handled in a responsible manner, with integrity, and there have to be rules and guidelines, this is a case that should matter to you.”

This isn’t an isolated concern. Separate reporting by Futurism has documented AI medical scribes hallucinating patient symptoms, and millions of Americans receiving dangerously flawed medical advice from AI tools used in place of physicians. What makes the Mayo allegations particularly striking is the institution’s scale — and its reputation.

When the number-one-ranked hospital in the country allegedly deploys a half-baked AI tool and punishes the person who said so out loud, it’s not just a Mayo story. It’s a signal for the entire industry.

What You Can Do Right Now

The lawsuit is still early. But the question it raises isn’t. It’s whether “world-class” and “cutting corners on AI safety” can coexist — and whether the answer is just “yes, apparently, if the competitive advantage is big enough.”

Here’s what you can do now:

1. Request your medical records from Mayo Clinic to review what tools and systems were documented in your care. Patients have a legal right to their records under HIPAA; submit a request directly to Mayo’s Health Information Management department.
2. Ask your care team directly whether AI tools were used in your diagnosis, treatment plan, or documentation — and request that any AI involvement be noted in your chart going forward. You are allowed to ask this. Ask it.
3. Monitor the lawsuit for class certification. The case is currently a whistleblower and wrongful termination suit, not a patient class action. If it expands to include patient claims, eligibility information will be filed with the court. Track it through the Minnesota state court docket or follow Minnesota Public Radio’s ongoing coverage.
4. Consult a medical malpractice attorney if you believe an AI error contributed to a misdiagnosis or delayed treatment at Mayo Clinic. Many offer free initial consultations and can advise whether your situation fits a potential claim.
5. File a complaint with the HHS Office for Civil Rights at hhs.gov/ocr/complaints if you believe your health data was mishandled by an AI system without your knowledge or consent.

If Mayo Clinic’s AI tools affected your care without your knowledge, report it to the HHS Office for Civil Rights at hhs.gov/ocr/complaints — or tell us what happened.

Written by: Companies Behaving Badly

The team behind it all.