Who's Responsible When AI Gets It Wrong? | Dr. Sarah Gebauer on Healthcare AI Governance

In this episode of MDAware: Upgrading Clinical Judgment, Dr. Graham Walker talks with Dr. Sarah Gebauer about what responsible AI in healthcare actually looks like — and what happens when the technology moves faster than our ability to govern it. A conversation about values, accountability, and why the physician-patient relationship still has to be at the center of it all.

Dr. Gebauer and Dr. Walker co-authored the MDCalc Physician's Charter in 2023, one of the earliest attempts by a group of practicing physicians to define a set of shared values for AI in healthcare. In this conversation, they revisit that document; what it got right, what it underestimated, and what a 2026 update would need to address. From the re-identification risks hiding inside "de-identified" data, to the tension between building accurate AI and protecting patient privacy, to the question of whether a sycophantic model is ever the right tool for a clinical setting, the conversation covers ground that most AI discussions in medicine still aren't touching.

Dr. Walker and Dr. Gebauer also look at where healthcare institutions are making real progress on AI governance, why rural hospitals may be further ahead than most people assume, and what it actually means to roll out a new tool responsibly — in medicine or anywhere else.

What You'll Learn

• Why HIPAA's definition of de-identified data may no longer be adequate in an era when AI can re-identify patients from surprisingly little information
• How the tension between diverse training data and patient privacy creates an ethical challenge the field hasn't fully resolved
• What questions healthcare systems should be asking before adopting any new AI tool, and why workflow fit matters as much as clinical accuracy
• Why AI sycophancy, models trained to be agreeable, could be a meaningful problem in clinical and patient-facing settings
• What a 2026 update to the Physician's Charter might need to address, including AI agents, foundation models, and tools that operate entirely outside the physician-patient relationship

Links & Resources

MDCalc: mdcalc.com

Physician's Charter: https://physicianscharter.ai/

Follow MDCalc on LinkedIn: linkedin.com/company/mdcalc

Follow MDCalc on X/Twitter: @MDCalc