Physician in the Loop

Physician groups reject claim AI is better informed than doctors; FTC probes OpenAI, Anthropic; Pennsylvania House passes health AI bill; Alabama law takes effect

Show Notes

Physician groups reject claim AI is better informed than doctors; FTC probes OpenAI, Anthropic; Pennsylvania House passes health AI bill; Alabama law takes effect

The voices on this episode are generated, and so is the script, from the day's post at Physician in the Loop. Every fact in the post links to its source.

Read the post and its sources

Six national physician groups said Sept. 30 that claims artificial intelligence is better informed than physicians diminish physician expertise and risk undermining patients' trust, after the health secretary praised AI second opinions at a summit.

The American Medical Association, the American College of Surgeons, the American College of Obstetricians and Gynecologists and three other groups issued the joint statement after Health and Human Services Secretary Robert F. Kennedy Jr. told the Make America Healthy Again Summit that AI offers a second opinion "much better informed than any doctor in the country," Medical Economics reported. The Federal Trade Commission confirmed an investigation of OpenAI, Anthropic and other AI companies over possible consumer harms.

In the states, the Pennsylvania House passed a bill requiring hospitals and physician offices to disclose AI use to patients and insurers to have a clinician review AI-assisted denials, and an Alabama law requiring a physician or other clinician to make prior authorization denials took effect Oct. 1. AdventHealth opened OpenEvidence to its clinical staff, and HHS launched four programs to bring AI into clinical trials.

In this episode

What is Physician in the Loop?

How technological innovation is changing medicine for patients and the people who practice on the front lines.

Nora: This is Physician in the Loop, the daily briefing on AI in medicine. I'm Nora Davis.

James: And I'm James Hudson. It's Friday, October 2nd, 2026.

Nora: Six national physician groups put out a joint statement Tuesday saying claims that AI is better informed than doctors diminish physician expertise and risk undermining patients' trust.

James: That sounds like a direct response to something. Who are they actually answering?

Nora: They name no one. But Medical Economics reported it followed Secretary Kennedy's remarks at the Make America Healthy Again Summit, where he said AI gives a second opinion, in his words, “much better informed than any doctor in the country.”

James: We talked about that quote already. What's new today is the physicians answering back.

Nora: Also today: the FTC confirming an investigation into OpenAI and Anthropic, Pennsylvania's House passing a health AI bill, and Alabama's prior authorization law taking effect. The six groups are the AMA, the Academy of Family Physicians, the Academy of Pediatrics, ACOG, the College of Physicians and the College of Surgeons.

James: Did Kennedy say anything else at that summit worth flagging?

Nora: Bloomberg reported he also said AI may “correct” experts on measures like vaccines and masks, and that Americans should be able to view their records on their phones and use AI to understand them.

James: So the physicians' statement is pushing back on the whole frame, not just the one line about second opinions.

Nora: Their own words: “Pitting physicians against technology or suggesting AI should replace physician expertise does not advance this important work.”

James: No names, but it's hard to read that as aimed anywhere else.

Nora: The Federal Trade Commission confirmed Tuesday it's investigating OpenAI, Anthropic and other AI companies over possible consumer harms.

James: Harms from what, specifically? Do we know?

Nora: ABC News reported the inquiry concerns possible unfair or deceptive practices, and that the agency could issue formal demands for company information. The FTC won't name the other companies, and there's no public document on the inquiry yet.

James: Anything concrete attached to this, or is it still just the confirmation?

Nora: Becker's noted Epic is patching security flaws, found in testing with Anthropic's Claude, that could have allowed unauthorized access to patient records. Separately, OpenAI extended three more tools to its ChatGPT for Healthcare customers on Tuesday.

James: So the regulator's looking at these companies the same week one of them is expanding its hospital product.

Nora: The Pennsylvania House passed House Bill 1925 Tuesday, 144 to 59.

James: What does it actually require, beyond the vote count?

Nora: Hospitals and physician offices using AI in clinical decision-making would have to tell patients in plain language, on their websites, and label AI-generated patient messages with a disclaimer and a way to reach a human. Insurers using AI in utilization review would need a clinician to review the patient's records and exercise judgment independent of any AI recommendation before denying, reducing or terminating benefits.

James: And if someone ignores that?

Nora: Penalties up to $10,000 per knowing violation, capped at $100,000 a year per entity. Representative Joe Hogan, one of the sponsors, said, “We must ensure that all final decisions are made by a human doctor.” It now heads to the Senate.

James: A follow-up on the physician-review trend: Alabama's law requiring a clinician to make AI-assisted prior authorization denials took effect Wednesday.

Nora: What's new beyond it taking effect?

James: Plans using AI for those determinations now have to base decisions on the enrollee's actual medical history and the clinical circumstances the provider presents, and certify annually to the state Department of Insurance that the AI doesn't rely on a group dataset and doesn't discriminate.

Nora: That certification piece is the new detail. Does it put any obligation on treating physicians themselves?

James: No. The law places no duty on them at all. It's entirely aimed at the plans.

Nora: AdventHealth opened OpenEvidence's clinical evidence search to its staff Wednesday, physicians, nurses, pharmacists, care managers and therapists, built right into their electronic health record workflows.

James: Any numbers on how many people or facilities that actually covers?

Nora: None. Becker's reported no count of users or facilities, and no results yet. Dr. Michael Cacciatore, the health system's chief clinical officer, said, “OpenEvidence gives our care teams a practical way to access current clinical evidence.”

James: And the health system is careful to say it's not replacing judgment, just the lookup.

Nora: Exactly their language: it supports clinician judgment, doesn't replace it.

James: UC San Diego Health, Harvard Medical School and Anthropic are building something a little different: a monitor that checks patient-facing AI advice after it's already live.

Nora: Checks it against what standard?

James: A “clinical constitution,” community-developed rules based on medical evidence and guidelines. It's part of the ADVOCATE program on AI-supported heart failure care, where Kaiser Permanente and Duke lead the real-world rollout.

Nora: We've heard about oversight tools being built alongside the thing they're meant to watch. Is that the idea here too?

James: Dr. Karandeep Singh, the health system's chief health AI officer, put it plainly: “In the long run, simply evaluating an AI tool before it goes live is not going to be enough.” He said the monitor would flag only true safety risks.

Nora: HHS and ARPA-H announced four programs Tuesday to modernize clinical trials.

James: Four programs is a lot. What's the lead one?

Nora: SURPASS, a phaseless trial design built on digital twins, a continuous inference engine reading data as it comes in, and an agentic layer that automates trial operations. The other three cover site activation, a national consent architecture, and patients' real-world data.

James: Any funding numbers attached?

Nora: None disclosed. STAT reported SURPASS is a five-year program taking submissions later this fall. The program manager, Daria Fedyukina, said, “Today's clinical trial system often requires too many stops, too much duplicated infrastructure, and too much time before researchers can understand whether a trial is on the right track.”

James: Before we go, something to watch. California's governor acted on those four health bills we mentioned, and there's now a fuller picture of the broader fight over AI and denials.

Nora: Newsom signed AB 1979 on clinician judgment, and vetoed AB 2575, which would have protected clinicians who override AI.

James: So the same split we flagged with the nurses' statement. And now Pennsylvania and Alabama are both pulling on that same thread, just from the insurer side.

Nora: Right, one on disclosure and clinician review, the other on certification against discrimination.

James: Three states, three different angles on the same basic worry: who's actually making the call when AI is in the room.

Nora: Sources and corrections are at physicianintheloop.org. I'm Nora Davis.

James: I'm James Hudson. See you tomorrow.