Date: July 21, 2026
Series: The Business of Health with Chip Kahn, Episode 13
As artificial intelligence (AI) transitions from a experimental novelty to a cornerstone of modern clinical practice, the healthcare industry finds itself at a critical crossroads. While the promise of accelerated diagnostics, personalized treatment plans, and optimized administrative workflows is immense, the legal and ethical framework governing these technologies remains largely in a state of flux.
In the latest installment of The Business of Health, host Chip Kahn sits down with Dr. Michelle Mello—a preeminent scholar in law and health policy at Stanford University—to dissect the "rules of the road" for AI. As the co-leader of Stanford’s Healthcare Ethical Assessment Lab for AI (HEAL-AI), Dr. Mello offers a sobering look at the challenges of accountability, transparency, and patient safety in an age of machine learning.
Main Facts: The Intersection of Innovation and Accountability
The rapid integration of AI into hospitals and clinics is outpacing existing regulatory structures. At the heart of this disruption lies a fundamental question: Who is responsible when an algorithm errs?
Dr. Mello argues that the current legal landscape, built largely on concepts of human-centered medical malpractice, is ill-equipped to handle the nuance of algorithmic decision-making. Key issues include:
- The Black Box Problem: Many AI diagnostic tools utilize deep learning models whose decision-making paths are opaque even to their developers. This lack of "explainability" makes it difficult for clinicians to verify the accuracy of a machine’s recommendation.
- Liability Shifts: When a physician relies on an AI suggestion that results in a patient injury, does the liability rest with the doctor, the software developer, or the hospital system that integrated the tool?
- Data Integrity and Bias: AI models are only as good as the data they are trained on. Without rigorous validation, models may perpetuate historical biases, leading to disparities in care for underrepresented patient populations.
Chronology: The Evolution of AI in Clinical Settings
To understand the urgency of the current regulatory debate, one must look at the rapid maturation of health AI over the past decade:
- 2018–2020: The Era of Pilot Programs. AI was largely relegated to radiology and dermatology, acting as a secondary "check" for human clinicians. Regulatory oversight was minimal, often categorized under general medical software guidelines.
- 2021–2023: Generative AI Emergence. The explosion of large language models (LLMs) shifted the landscape. AI began to assist in patient communication, clinical note-taking, and complex triaging, moving beyond simple image recognition.
- 2024–2025: Institutional Integration. Large health systems began adopting proprietary AI models for predictive analytics, such as forecasting patient deterioration or managing hospital capacity. This created a demand for standardized vetting processes.
- 2026: The Regulatory Catch-Up. As of mid-2026, the focus has shifted from "can we build it?" to "how do we govern it?" Organizations like HEAL-AI are now at the forefront of creating the "ethical safety checks" that policymakers are struggling to codify.
Supporting Data: The Scale of the Challenge
According to research cited by Dr. Mello and her colleagues at Stanford, the scale of AI implementation in the United States is staggering. Recent industry surveys indicate that:
- Deployment Rates: Over 65% of major academic medical centers in the U.S. have deployed at least three distinct AI tools for clinical decision support.
- The Error Gap: Studies on diagnostic AI tools have shown a variance in performance across different demographic groups, with some models exhibiting a 15–20% decrease in accuracy when applied to patients from non-white demographics due to training data imbalances.
- Policy Deficit: A 2026 analysis of state and federal medical board regulations found that fewer than 10% of jurisdictions have specific guidelines regarding the use of "black box" algorithms in patient care, leaving most practitioners operating in a regulatory gray area.
Official Responses and Ethical Frameworks
Dr. Mello, through her work at Stanford’s HEAL-AI, advocates for a "continuous monitoring" approach rather than a one-time regulatory approval.
"The traditional FDA approval process for medical devices—which often focuses on a static snapshot of a product—is not sufficient for AI," Mello explains. "AI models learn and evolve. A tool that is safe today may be biased or inaccurate tomorrow as it encounters new, unforeseen patient populations."
The HEAL-AI Protocol
The lab’s approach involves a rigorous, three-pronged assessment:
- Technical Validation: Ensuring the model meets performance benchmarks in local clinical environments.
- Ethical Review: Evaluating the potential for systemic bias and ensuring the model does not violate patient autonomy.
- Human-in-the-loop Implementation: Establishing strict protocols that ensure a qualified physician remains the final decision-maker, mitigating the risk of "automation bias"—where clinicians over-rely on computer suggestions.
Implications: The Future of Liability and Care
The implications for the business of healthcare are profound. As Dr. Mello notes, the industry is entering an era where insurance policies for medical malpractice will need to be entirely restructured.
Legal and Insurance Shifts
If hospitals are to be held liable for the performance of their AI tools, they will need to implement comprehensive insurance riders specifically for "algorithmic injury." This could significantly increase the cost of adopting new technology, potentially favoring large hospital networks over smaller, independent practices that cannot afford the necessary risk mitigation infrastructure.
The Patient-Provider Relationship
Perhaps the most significant implication is the effect on the human element of medicine. The danger, according to Mello, is that AI could reduce the patient to a data point. If the clinical decision-making process becomes too automated, the compassionate, nuanced judgment of a human provider—who can see the patient’s socio-economic, emotional, and cultural context—could be lost.
Policy Recommendations
Dr. Mello suggests that policymakers move toward a model of "co-regulation." This would involve:
- Mandatory Transparency: Requiring developers to disclose the data sources and limitations of their algorithms to the hospitals that purchase them.
- National AI Registries: Creating a central database where adverse events involving AI are tracked, allowing for faster identification of systemic flaws.
- Professional Certification: Developing new training modules for medical professionals to ensure they understand how to interpret AI outputs and identify when an algorithm is hallucinating or malfunctioning.
Conclusion: A Collaborative Path Forward
As the Business of Health series continues to examine the evolution of AI, it is clear that technology is not a panacea for the systemic issues facing healthcare. Rather, it is a powerful tool that, if mismanaged, could exacerbate existing inequities and create new, complex legal vulnerabilities.
Dr. Mello’s work serves as a reminder that the "rules of the road" for the digital age of medicine are not merely a technical concern; they are a moral imperative. As AI continues to race into our hospitals, the ultimate goal must remain the same as it has always been: the safety and well-being of the patient. The challenge for the coming years will be to ensure that our laws and ethics move at the same speed as our code.
About the Host:
Charles N. Kahn III is a senior visiting fellow at KFF, a visiting senior fellow at the American Enterprise Institute, and a nonresident senior scholar at the University of Southern California’s Schaeffer Center for Health Policy & Economics. He serves as co-chair of the international Future of Health collaborative.
About the Guest:
Michelle Mello, JD, Ph.D., is a Professor of Law at Stanford Law School and a Professor of Health Policy at the Stanford University School of Medicine. Her work focuses on the intersection of law, ethics, and health policy, with a recent emphasis on the regulatory hurdles of artificial intelligence.
About the Author
