AMA says AI cannot be licensed as a doctor and vendors should own its errors

• WatchRegulation Nov 10, 2026
  AMA delegates vote that AI cannot hold a medical license or a physician's title, and that vendors should be liable for defective tools

  The AMA said it will seek model legislation separating technology authorization from licensure, and state and federal laws making vendors liable for defective AI.

  All physicians
  The American Medical Association's House of Delegates adopted policy Nov. 9 at its Interim Meeting in Orlando, Fla., stating that AI cannot receive physician licensure, professional titles or certification to practice medicine, and that AI tools belong in physician-led teams under the treating physician's direction. When a physician uses a clinical AI product as intended and exercises appropriate judgment, the policy places liability for a defective product's errors on the vendor, manufacturer or distributor, and when a hospital, health system or payer mandates a specific tool, on that entity and the vendor together. It also calls for liability protection for physicians who decline to use an AI tool or follow its recommendation, product liability insurance as a condition of FDA authorization, disclosure of known failure rates and a ban on contract terms that shift liability to physicians. "Patients deserve to know who is responsible for their care," AMA President Dr. Willie Underwood III said. The policy builds on the liability framework the AMA released Oct. 29; a separate resolution calls for physician-led oversight of AI used in credentialing, privileging and peer review.
  Sources: American Medical Association, AMA policy summary, AMA liability framework (Oct. 29), Becker's Hospital Review
• FYIEvidence Nov 9, 2026
  Patients are more comfortable with AI diagnosis than their doctors think, with a 20-point gap when a clinician can override it
  Primary care
  Fifty-six percent of patients said they would be comfortable with AI making diagnosis and treatment decisions that a provider can override, while primary care physicians estimated 36%, according to a West Health-Gallup survey released Nov. 9. The probability-based patient survey of 5,330 adults who had seen a primary care provider in the past year ran Aug. 12 to 24; the opt-in physician survey of 1,007 primary care physicians ran Aug. 7 to 23. Comfort fell without clinician review: 9% of patients accepted AI diagnosis and treatment decisions with no provider review, and physicians guessed 6%. Becker's reported that 89% of the physicians had used AI for work in the past year and 20% had received formal training.
  Sources: Gallup, Becker's Hospital Review
• FYIMoney Nov 10, 2026
  Oura agrees to buy Counsel Health, whose AI chats escalate to physicians who can diagnose and prescribe, for $625 million in cash
  Primary carePractices
  Smart-ring maker Oura signed a definitive agreement Nov. 10 to acquire Counsel Health, an AI-native primary care company, for $625 million in cash, with closing expected before the end of the calendar year, the companies said. Counsel's AI draws on medical records, guidelines and continuous biometric data and can escalate a chat to a board-certified Counsel physician within minutes who can provide diagnoses, prescriptions and referrals; Counsel says it replaces nurse triage lines and telehealth for health plans, health systems and employers. Oura says it has more than 6 million members. The release reports satisfaction figures, including a 4.6 out of 5 rating from members who saw a Counsel physician, and no clinical outcome data. Counsel founder Dr. Muthu Alagappan will become Oura's chief clinical AI officer, TechCrunch reported.
  Sources: Oura and Counsel Health (Business Wire), TechCrunch, STAT
• ActIncident Nov 11, 2026
  False alarms on 80,938 Tandem Mobi insulin pumps can halt insulin delivery, in a top-tier recall tied to 230 adverse events

  Tandem tells users to answer every occlusion alarm, keep backup insulin ready and call support if alarms recur; pumps with frequent false alarms are exchanged.

  EndocrinologyPrimary care
  The FDA classified Tandem Diabetes Care's correction of its Mobi insulin pumps as a Class I recall, the most serious type, after friction in the drive module was found to cause false occlusion alarms that suspend insulin delivery until the user responds. The correction covers 80,938 pumps, 77,559 of them in the United States, and all models and serial numbers may be affected, according to the FDA's recall entry posted Oct. 30. Tandem had received 230 reports of adverse events requiring medical intervention or hospitalization as of Oct. 29, a company spokesperson told MedTech Dive, which reported the figures Nov. 11 and said no deaths had been reported. Users may keep using the pumps under the company's instructions.
  Sources: FDA recall database, MedTech Dive
• FYIEvidence Nov 12, 2026
  Aspirin fails to prove noninferior to rivaroxaban or warfarin for clot prevention after hip or knee replacement in an 18,883-patient trial
  Hospital medicineOrthopedic surgery
  Low-dose aspirin did not meet noninferiority against rivaroxaban or warfarin for preventing death or symptomatic venous thromboembolism after elective hip or knee arthroplasty, according to the PEPPER comparative-effectiveness trial published in the New England Journal of Medicine. Patients received aspirin 81 mg twice daily, warfarin to an INR of 2.0 or rivaroxaban 10 mg daily for 30 days; in the per-protocol analysis of 15,317 patients, the 6-month primary outcome occurred in 1.26% with aspirin, 1.23% with warfarin and 0.94% with rivaroxaban, against a noninferiority margin of 0.69 percentage points. Bleeding, wound reoperation or myocardial infarction occurred in 2.91%, 3.36% and 3.48%, and neither anticoagulant was noninferior to aspirin for safety. The Patient-Centered Outcomes Research Institute funded the trial.
  Sources: New England Journal of Medicine, PubMed
• FYIEvidence Nov 11, 2026
  Holding ACE inhibitors or ARBs the morning of surgery cuts perioperative hypotension to 31% from 46% in older patients
  Hospital medicineAnesthesiology
  Withholding an ACE inhibitor or angiotensin receptor blocker on the morning of noncardiac surgery lowered perioperative hypotension without raising 30-day complications or mortality, according to a double-blind, placebo-controlled trial of 2,036 patients ages 60 to 80 at five Chinese teaching hospitals, published in The BMJ. Hypotension, defined as mean arterial pressure below 65 mm Hg or a fall of 30% or more requiring treatment, occurred in 31.2% of the placebo group and 45.5% of those who took their usual dose, an absolute difference of 14.4 percentage points (adjusted odds ratio 0.52). Vasopressor use also fell, while intraoperative fluid volume, postoperative hypotension, neurocognitive disorders and major cardiocerebral events did not differ. The authors described the results as high-certainty evidence for preoperative drug management in this population.
  Sources: The BMJ, PubMed