Daily scan · 2026-06-02

Microsoft launches Copilot Health for consumers with no BAA and no HIPAA backstop, leaving patient-supplied lab results and wearables data outside US health-privacy law.

Florida AG Uthmeier: First state to sue OpenAI and Altman, alleging ChatGPT addiction, behavioral harm, and direct role in a teen suicide and the FSU shooting plot.

Nursing researchers propose a human-centered framework for empathic and agentic AI to guide cancer patients through care navigation, with nurses owning the oversight layer.

Clinicians edit ambient AI scribe drafts to remove hedging language before saving notes, shifting epistemic certainty in ways patients reading the chart never see.

Authors argue AI in clinical encounters redistributes voice and responsibility between patient and clinician, and propose accountability tests for relationship-centred care.

Paper maps how algorithmic outputs are quietly absorbed into the clinical standard of care, raising liability and patient-consent questions when AI advice contradicts a physician.

Experimental study documents AI-conformity, where exposure to chatbot answers shifts human moral and clinical decisions toward the model's framing within minutes.

Frontline clinician's letter from Arizona documents uninsured patients using ChatGPT as a primary-care substitute and asks who owns the resulting risk.

Digital-health vendors target the ACCESS gap with patient-facing AI navigators, betting that LLM intake replaces the front desk before it replaces the clinician.

Argues that virtual-first care and AI triage are dissolving the physical waiting room, pushing intake and pre-visit reasoning onto the patient and the model.

States revive corporate-practice-of-medicine laws to rein in telehealth-plus-AI prescribers, threatening patient access to online weight-loss and mental-health care.

Opinion piece argues the AI industry's overpromising and underwarning on consumer health features is the proximate cause of the new public-trust gap.

Arizona physician describes uninsured patients arriving with ChatGPT printouts as a stand-in for primary care the system no longer offers them.

UPI: Reports OpenAI's o1 hit 78% accuracy on complex NEJM cases and beat ER physicians, while clinicians still outperform on treatment selection and patient management.