U.S. Health Officials Accelerate Medical A.I. Rollout for Diagnosis and Prescribing as Safety Evidence Lags
Updated
Updated · The New York Times · Sep 14
U.S. Health Officials Accelerate Medical A.I. Rollout for Diagnosis and Prescribing as Safety Evidence Lags
3 articles · Updated · The New York Times · Sep 14
Summary
The Trump administration is pushing A.I. deeper into U.S. care, backing projects that would let software agents diagnose patients and prescribe treatments.
HHS officials and others inside government worry the shift is moving too fast, with too little proof that the tools are safe and effective for direct patient care.
Vinod Khosla has emerged as an influential voice in talks with top health officials, including CMS chief Mehmet Oz, even as Khosla backs Curai Health, a company run by his son.
Chris Klomp, awaiting a confirmation hearing this week to become HHS's No. 2 official, called A.I. in patient care the agency's "holy grail" for improving health and cutting costs.
The push comes as debate over A.I. safety is intensifying more broadly, and as medicine still uses the technology mainly for administrative help rather than replacing physicians' clinical roles.
Why are health officials embracing agentic medical AI when patients, doctors, and regulators still question bias, safety, and accountability?
Can AI diagnose and treat patients safely before real-world evidence catches up to federal enthusiasm?
Medical AI in 2026: Record Physician Adoption, Shadow AI Threats, and the Race for Trust and Equity
Overview
In 2026, U.S. healthcare saw a dramatic surge in AI adoption, driven by physician burnout and heavy administrative workloads. Hospitals rapidly deployed ambient AI scribing tools, which saved clinicians significant time and improved well-being, but also introduced new risks like AI-generated errors for which clinicians remain legally responsible. Slow IT approval processes led many staff to use unauthorized 'Shadow AI' tools, causing serious data security and HIPAA concerns. Meanwhile, many AI medical devices entered the market with little clinical validation, resulting in high recall rates, especially among investor-driven companies. Regulators are now shifting toward continuous monitoring and competency-based evaluation to address these evolving challenges.