Daily Update for September 25, 2026

Show Notes

Blue Cross ties $942 million to hospital AI coding; Oz says AI will raise costs first; clinicians catch 16% of AI hallucinations; Cigna and OpenAI start with cancer

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

Hospitals' AI coding tools added $942 million to Blue Cross Blue Shield plans' inpatient costs from early 2023 through 2025 without matching changes in treatment, the Blue Cross Blue Shield Association said Sept. 24.

The association found that the share of inpatient claims coded as medically complex rose from about 37% to about 40% while hospitals with the most complex coding delivered similar or less intensive care, Fierce Healthcare reported. The American Hospital Association said inpatients are older and sicker. The same day, Centers for Medicare & Medicaid Services Administrator Dr. Mehmet Oz told an Oracle conference in Orlando that AI will be "inflationary" in the short term because it will "turbocharge" billing systems. A multicenter study in npj Digital Medicine found that junior clinicians identified 15.8% of hallucinations in GPT-4o outputs.

Two announcements this week aimed AI at cancer care. The Cigna Group and OpenAI said their first joint tools will serve oncology nurses and case managers, and Oracle Health announced an oncology electronic health record with AI agents but gave no launch date. Cancer was the most-searched disease area on Medscape's AI search tool in all nine countries the company studied in the first half of 2026, according to Medscape data published through Stacker.

In this episode

• Blue Cross group says hospital AI coding added $942 million in inpatient costs over two years with no matching rise in treatment
• Oz says AI will be "inflationary" in the short term and points to accountable care organizations as the fix
• Junior clinicians identify 15.8% of GPT-4o hallucinations in multicenter simulation; 13.1% catch none
• Cigna and OpenAI build AI tools for oncology nurses and case managers first; insurer says models make no clinical or coverage decisions
• Oracle Health announces oncology EHR with AI agents and lifetime chemotherapy dose tracking; no customers or launch date
• Neko Health opens first U.S. clinic in Manhattan with $499 scan that includes 6,000-image skin exam
• August AI, whose chatbot claims 9 million users worldwide, launches $39-a-month U.S. membership with one physician visit
• Mass General Brigham says 77% of patients using its AI intake agent see a doctor the same day; panel splits on telling patients about AI

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, September 25th, 2026.

Nora: Blue Cross Blue Shield's own trade group says hospital AI coding tools added $942 million to its plans' inpatient costs in just two years, from early 2023 through 2025, with no matching rise in treatment.

James: $942 million with no matching change in care. Hospitals say patients are just sicker. Two very different explanations for one number.

Nora: I'm leading with that Blue Cross analysis, then I've got a new study on how often clinicians actually catch AI hallucinations.

James: I've got CMS Administrator Dr. Oz calling AI inflationary in the short term, and the first wave of AI tools built specifically for cancer care.

Nora: So, the Blue Cross Blue Shield Association represents 31 independent plans covering more than 100 million people. It says $653 million of that $942 million came from secondary diagnoses.

James: What did those diagnoses do exactly?

Nora: They moved about 55,000 additional cases into higher-paying diagnosis groups, at an average of about $11,000 per case. Fierce Healthcare had the numbers.

James: So more codes, more money. Did the sicker patients actually get sicker care?

Nora: According to the association, no. Hospitals with the highest coding complexity showed similar or lower ICU use, transfusions, reoperations and length of stay than their peers.

James: Any specifics?

Nora: Becker's reported top-quartile hospitals diagnosed anemia 38% more often, but transfused less, 16.9% against 19.3%.

James: That's an odd pair of numbers.

Nora: The association's Dr. Razia Hashmi put it plainly: 'The likelihood that this is technology-enabled upcoding is higher, in my view.'

James: And the hospital side pushes back?

Nora: The American Hospital Association says inpatients are older and sicker, and points to a roughly 5% rise in case-mix index from 2019 to 2024, according to Medical Daily.

James: So who's right?

Nora: The sources don't settle that. PYMNTS reported the analysis names ambient scribes among the tools surfacing these secondary diagnoses. Becker's says hospital leaders feel payers have automated claim review and denials at a scale hospitals can't match.

James: Same day, CMS Administrator Dr. Oz told an Oracle health and life sciences summit in Orlando that AI is going to be inflationary in the short term.

Nora: His words?

James: 'Short term, AI is going to be inflationary because it's going to turbocharge the ability of the current billing systems to work more effectively.' Healthcare Dive had it.

Nora: So his own regulator is naming the same problem we just described. What's his fix?

James: He pointed to accountable care organizations, paid on outcomes instead of fee for service. Healthcare Dive reported that would reward using AI for clinical efficiency rather than billing.

Nora: Does he think it's worth the tradeoff?

James: He's blunt about it: 'I guarantee you we will lose lives if we don't use AI in the day-to-day trench warfare of fighting disease in America.'

Nora: So billing gets worse before care gets better, in his telling.

James: That's his bet, yes.

Nora: Next, how much can we actually trust clinicians to catch AI's mistakes. A multicenter study in npj Digital Medicine found junior clinicians identified just 15.8% of hallucinations in GPT-4o outputs.

James: 15.8%. What happened to the rest?

Nora: The study, led by Yunyun Zhang and colleagues, had junior clinicians at several centers review GPT-4o output in simulated clinical scenarios. 13.1% caught none at all.

James: Did detection at least improve on the riskier cases?

Nora: No. Detection didn't improve as the clinical risk of a scenario went up. Most of the variation came from differences between clinicians, not between scenarios.

James: So it's less about the case and more about who's reading it.

Nora: The authors wrote that current safeguards rely on a 'clinician-in-the-loop' model that 'assumes that clinicians can reliably identify and correct these hallucinations.' They don't think that assumption holds.

James: What do they want instead?

Nora: Structured human-AI workflows and tiered clinical certification. Worth saying, this was simulated cases, not live patient care.

James: Two announcements this week both aimed AI squarely at cancer care. The Cigna Group and OpenAI said their first joint tools will serve oncology nurses and case managers, not patients directly.

Nora: What do the nurses actually get?

James: A single view of clinical, pharmacy, behavioral health and benefits data, plus patient-reported side effects and questions between visits, for nurses and case managers at Cigna Healthcare and Accredo Specialty Pharmacy.

Nora: Who's actually making the decisions?

James: Cigna says clinicians keep that responsibility, and, quote, 'no patient or member data is maintained by OpenAI or used by OpenAI for model training.'

Nora: No launch date, I assume.

James: None. No patient count or financial terms either, Fierce Healthcare reported.

Nora: Oracle Health had a cancer announcement of its own.

James: An oncology electronic health record with AI agents and chemotherapy dose tracking. Seema Verma said it's 'designed to bring the patient's story together and provide personalized treatment recommendations.'

Nora: But?

James: Health System CIO reported the release calls this 'general product direction' and 'not a commitment to deliver.' No customers, no launch date.

Nora: Cancer's clearly where the attention is. Medscape's AI search tool data, published through Stacker, found cancer was the most-searched disease area in all nine countries the company studied in the first half of 2026.

James: So the clinical demand is real even if the products aren't out the door yet.

Nora: Before we go, two things to watch.

James: First, payer AI denials. Connecticut bars AI-only denials on its state plans starting January 1, 2027.

Nora: That's a hard line drawn right under this week's Blue Cross number.

James: California's got two related bills, AB 1979 and AB 2575, awaiting action by September 30.

Nora: Second, the agent layer moving onto the wards. The line everyone's watching for hasn't moved: an AI agent placing an order in a US hospital without a signature.

James: Has that happened yet?

Nora: Not that's been reported. It's why a kill switch is now a design requirement at Parkview and Brigham.

James: And why Mass General Brigham says it's 'not allowing autonomous AI across different platforms.'

Nora: So the guardrails are catching up before the incident, not after.

James: That's the bet, that a checklist item now beats a headline later.

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

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