3, 2, 1: Health AI Brief
Every Friday
December 26, 2025

AI is reshaping healthcare fast. Below are 3 key AI developments, 2 studies, and 1 takeaway to help you better lead with AI. Target read time: 5 minutes.

3 Market Signals
Trump administration cancels AI "model cards" rule

The administration eliminated Biden-era requirements that software vendors disclose how AI clinical decision support tools are developed and evaluated. ASTP/ONC stated there is "no publicly available evidence" that transparency requirements led to positive patient outcomes. The changes take effect December 29.

So what?

Health plans and health systems now bear more of the AI vetting burden—without standardized disclosures to compare products or assess fairness across patient populations.

Read the full story →

Mass General Brigham launches AI-powered virtual primary care

Care Connect pairs AI chatbots with remote physicians for 24/7 virtual primary care. Patients spend about 10 minutes chatting with an AI agent before a video consultation. The program currently sees 40-50 patients daily with 12 remote physicians, and will expand to Massachusetts and New Hampshire residents by February.

So what?

Novel combination of AI + PCP working in the real world. Yet, not without critics: undermines comprehensive, in-person care. Experiment continues regardless, AI tailwinds too strong.

Read the full story →

HHS seeks input on AI adoption in clinical care

HHS issued a Request for Information asking stakeholders how to structure reimbursement models that incentivize AI adoption, enhance interoperability, and address HIPAA compliance. The RFI complements HHS's broader AI Strategic Plan. Comments are due January 19, 2026.

So what?

This shapes the regulatory runway for 2026 and beyond. If your organization has a position on AI reimbursement, now is the time to submit it.

Read the full story →

2 Research Studies
NEJM AI: Verifying facts in LLM-generated patient care documents

Researchers developed methods to fact-check clinical documentation generated by large language models using electronic health record data. The study addresses a critical gap as LLMs increasingly draft patient notes, discharge summaries, and clinical communications.

Why it matters

As LLM-generated clinical docs proliferate, verification infrastructure becomes table stakes for safe deployment. Manual oversight is ideal, but not sustainable.

Read the study →

UC Davis: How patients feel about AI transcription

A survey of nearly 1,900 patients found 48% view AI scribes as beneficial, 33% were neutral, and 19% expressed concerns. Top worries: note accuracy (39%), privacy (13%), and being recorded (13%). Notably, younger patients (18-30) were more skeptical than older patients.

Why it matters

Patient consent and transparency matter. Early notification—at scheduling or check-in—is what patients want.

Read the study →

1 Key Insight
The belief-action gap: 80% see AI value, only 33% using at scale

The first line of Deloitte's 2026 Healthcare Outlook says it all: "More than two-thirds of leaders from health plans and health systems anticipate outperforming their competitors in 2026." Objectively, that's not possible—yet it's a perfect example of the gap between belief and action.

The data tells a similar story: Over 80% believe generative AI can deliver "moderate-to-significant value" in 2026—but only one-third are using AI at scale. The rest? 49% are still experimenting. 18% haven't adopted AI at all.

The cited brake: policy and regulatory uncertainty, with 43% of executives feeling "uncertain" or "neutral" about the industry's near-term outlook—up from 28% last year.

Takeaway

Regulatory clarity is definitely lagging, but early movers are building the institutional muscle that everyone else will have to develop later. When clarity does ultimately arrive, those still waiting will find themselves playing catch-up.

Read the Deloitte report →

Know someone who'd find this useful?

Forward to a Colleague
HealthLeader.AI

Signal over noise. Every Friday.

Archive Preferences Unsubscribe

You're receiving this because you subscribed at healthleader.ai
HealthLeader.AI © 2025