Voice 1: This is Physician in the Loop, the daily briefing on AI in medicine. It's Friday, November 20th, 2026.
Voice 1: Researchers said AI helped them find a critical flaw in Java in about an hour, and build a working exploit in two more, for less than $200 in AI tokens. Hospitals are racing to patch it now.
Voice 2: An hour to find it, two hours to weaponize it. What's actually at risk if a hospital misses this one?
Voice 1: The flaw, called Java4Shell, lets an attacker run code remotely without even logging in. It sits in the Base64 decoder of Java's HotSpot virtual machine. The Cybersecurity and Infrastructure Security Agency urged organizations to patch it on November 17th, and Oracle released a fix that very same day.
Voice 2: What should hospital IT teams actually be watching for, beyond just installing the patch?
Voice 1: CISA says to look for Java bundled inside third-party apps, container images and appliances, and to watch for unexpected JVM crashes or rogue runtimes.
Voice 2: Same-day fix is good news. Has anyone said whether hospitals have actually been hit?
Voice 1: The American Hospital Association said it knows of no effect on hospital operations, beyond the rush to deploy defenses. Epic said its own systems weren't affected.
Voice 2: That's a lot of hiding places for one bug, even with a same-day patch.
Voice 1: John Riggi, the American Hospital Association's national cybersecurity adviser, said hospital cyber teams are trying to “patch before they hack.”
Voice 2: A study out today turns from hospital networks to children getting head CT scans. A cohort of more than 3.7 million children tied the scans to a measurable rise in brain tumors.
Voice 1: How big a rise are we talking about?
Voice 2: About one extra brain tumor for every 1,266 head CT scans, according to JAMA. The cohort recorded 1,536 brain tumors over more than 45 million person-years. Risk rose with cumulative radiation dose, and a single head CT delivered an average of about 66 mGy to the brain.
Voice 1: Did that risk keep climbing the more imaging a child had, or level off?
Voice 2: It kept climbing. Hazard ratios compared with no exposure went from under 2 at the lowest doses to nearly 4 at the highest. By age 21, exposure like one head CT was tied to about 8 extra brain tumors per 10,000 children, and the researchers estimated imaging caused about 10% of the tumors in the study.
Voice 1: Is that an argument for skipping scans a child actually needs?
Voice 2: The authors didn't say that. They said it supports careful clinical justification, and keeping the dose as low as possible when imaging with radiation is actually called for.
Voice 1: That's a number pediatricians will be sitting with for a while.
Voice 2: Context for the next order, not a reason to skip one that's actually needed.
Voice 1: The FDA had its own warning this week, on a device already sitting in patients: a needle-free IV valve that can stay open after it's disconnected.
Voice 2: Open how? What's actually happening inside it?
Voice 1: B. Braun's Caresite Micro Luer Access Device is used for blood draws, IV therapy and medications. After disconnection, the internal piston can stay depressed instead of closing, so the valve stays open.
Voice 2: And what can an open valve like that actually cause?
Voice 1: Backflow and leakage, and less often, air getting into the line. FDA listed the risks as blood loss, interrupted therapy, medication loss, air embolism and microbial contamination. Neonates, children, critically ill patients and anyone on life-sustaining therapy face the highest risk.
Voice 2: Has the company said what's actually wrong with the valve?
Voice 1: They traced it to missing silicone lubricant at the piston. B. Braun told customers in a letter on November 13th to inspect every device after disconnection, replace any that leak or stay open, and replace the whole extension set if the valve is bonded to it.
Voice 2: Any injuries reported so far?
Voice 1: As of November 3rd, the company had reported none: no serious injuries or deaths, FDA said.
Voice 2: A panel that hasn't met in over a year finally sat back down this week.
Voice 1: Which panel, and why the gap?
Voice 2: The U.S. Preventive Services Task Force, the group that decides which preventive services insurers have to cover. It convened on November 19th, its first meeting in more than a year, after Health Secretary Robert F. Kennedy Jr. canceled its meetings and overhauled its membership.
Voice 1: What's actually on the table after sitting idle that long?
Voice 2: Final recommendations on cervical cancer screening, screening and counseling for unhealthy alcohol use, and counseling to prevent perinatal depression. None of those have been updated in at least five years, and all three have been waiting for finalization since last year.
Voice 1: Anything beyond catching up on the backlog?
Voice 2: They also heard an evidence presentation on diet, physical activity and weight loss to prevent cardiovascular disease. The next day's session takes up chronic kidney disease, colorectal cancer and lung cancer screening.
Voice 1: A full agenda for a panel that went quiet for over a year.
Voice 2: Before we go, something to watch. A Vizient survey found that among hospitals reporting the biggest returns on AI, 17% had made AI-related layoffs.
Voice 1: Is anyone actually tying that to physician jobs?
Voice 2: No. There's still no verified case of a U.S. health system cutting physician positions because of AI. But Houston Methodist's chief innovation officer told Becker's the system now asks vendors what headcount reduction their tools have actually produced, and that time savings alone don't meet its bar.
Voice 1: That's a specific number to hold a vendor to, instead of just a promise.
Voice 2: Asking for the headcount number up front is a different kind of due diligence than most AI tools have had to pass.
Voice 1: Sources and corrections are at physicianintheloop.org, where you can sign up for the free email briefing. This briefing was read by AI voices.