Hospitals put clinical AI into their core software. Regulators draw lines around what it may do.

Northwestern Medicine says it has made OpenEvidence, an AI clinical decision-support service, available from within its Epic electronic health record across the system.¹

A JAMA study of verified users found that OpenEvidence received 271 million queries from 952,008 U.S. clinicians from January 2024 through June 2026.² Intermountain Health says about 75% of its radiologists actively use Microsoft's Smart Impression tool to draft the impression section of radiology reports.³

OpenEvidence's use from inside Epic and Smart Impression's use in radiology reporting show that clinical AI is moving into the systems where clinicians write notes, read records, and sign reports.¹³

The Peterson Health Technology Institute said AI used for documentation, coding, billing, and prior authorization can raise costs without improving patient outcomes under fee-for-service payment incentives.⁴

The American Hospital Association disputed a Blue Cross Blue Shield Association analysis that linked AI-enabled billing tools to $942 million in added coding costs, arguing that sicker and more complex patients explain the change.⁵

The Guardian reported that AI is increasingly taking notes in exam rooms, assigning codes, sending billing messages, and powering patient communications about medical debt as expected coverage cuts raise out-of-pocket exposure.⁶

The American Medical Association's new framework says an AI system's clinical authority should depend on its function, evidence, risk, oversight, and accountability, with stronger evidence for higher-risk and more autonomous tasks.⁷

The Federation of State Medical Boards has convened doctors, hospitals, insurers, technology companies, and state representatives to draft guidance for AI that performs clinical functions with limited or no direct physician supervision.⁸

The U.K.'s Medicines and Healthcare products Regulatory Agency said its 2030 strategy will update regulation for AI and personalized medicine while maintaining safety and effectiveness standards.⁹

Iambic Therapeutics raised $150 million in an initial public offering to finance clinical development of cancer medicines developed with AI at the center of its research approach.¹⁰

Researchers used AI protein-design methods to create enzymes designed to cut selected proteins, including proteins associated with neurological disease.¹¹

Google DeepMind says its Antigravity research workflow identified an antimicrobial-peptide candidate from protein databases, and that the lead is now undergoing wet-lab validation.¹²