Healthcare AI News 10/15/25
News

Google backtracks on an internal health policy that required employees to allow a third-party AI tool from startup Nayya review their personal data when enrolling in health benefits. Nayya’s tool asks health and lifestyle questions to recommend benefit options. The company says, “Our intent was not reflected in the language on our HR site.”
Business
OpenAI will enhance ChatGPT to allow Walmart customers to complete Instant Checkout purchases directly from the app. The partnership threatens traditional search engines by delivering predictive, contextual results from chats such as meal planning.

Duos, which offers an AI-driven platform that helps Medicare Advantage members manage social determinants of aging, raises $130 million in a strategic growth equity round. The system guides members through their plan benefits and services to ensure that they complete high-value actions that improve health plan revenue and member retention.
Research
A JAMA summit report says that AI is rapidly spreading across healthcare but lacks consistent evaluation, oversight, and real-world outcome data. It calls for coordinated frameworks for testing, monitoring, and incentivizing safe, effective AI use across clinical and operational settings. The report also flags legal uncertainty, noting that failure to use AI could be seen as a breach of standard of care, yet malpractice liability may be difficult to assign among clinicians, health systems, and AI developers.
Researchers propose a framework of “confidence calibration and transparency” to improve clinician trust in AI by scoring its recommendations. Clinician override of AI recommendations dropped from 33% to less than 2% when the AI scored itself on its level of confidence, transparency, and semantic similarity to a clinician-verified diagnosis.

Researchers introduce an AI system that presents Grand Rounds-type medical case information and the logic its used to reach a diagnosis. The authors posted the AI’s output, which is generated in both written and narrated slide-based presentation form, alongside an expert clinician’s version. The AI-created video is startlingly insightful.
Other

An AI engineer who probes AI tools for undocumented features finds references to a “clinician mode” within ChatGPT’s web app code.
Contacts
Mr. H, Lorre, Jenn, Dr. Jayne.
Get HIStalk updates.
Send news or rumors.
Follow on X, Bluesky, and LinkedIn.
Sponsorship information.
Contact us.

I dont think anything will change until Dr Jayne and others take my approach of naming names, including how much…