Weekly AI and Healthcare Policy Stories
We run a daily intelligence filter on clinical AI and healthcare policy so you don't have to wade through vendor press releases. Here are the two stories from this week that actually affect independent patient advocacy. They can be summarized as: 52% of U.S. physicians now use AI daily in clinical practice, but 75% of them say they caught a factual inaccuracy from those tools over the past year.
1. The AMA drew a hard statutory line on AI «practicing» medicine. In its latest Advocacy Update, the AMA adopted formal policy stating that AI cannot be licensed as a physician, hold clinical privileges, or replace physician-led care. More importantly for our cases: the AMA is pushing a liability framework where health systems and vendors, not just the individual doctor, bear responsibility when a hospital-mandated AI tool causes patient harm. See the link in the comments.
2. Clinical AI adoption is outpacing verification. According to Doximity’s new report (covered by Healthcare Dive), daily physician AI use jumped from 37% to 52% in a single year, moving well beyond ambient scribing into clinical decision support and literature lookups. Yet a third of those clinicians admit they do not verify AI-generated clinical info every time. Yikes! See the link in the comments.
When a treatment plan shifts unexpectedly or a chart summary contains an unverified assumption, don't assume a human clinician generated it from scratch. Ask directly: «Was an ambient scribe or clinical decision tool used to generate this note or recommendation, and has the attending reviewed the underlying chart?»
Have you spotted an AI-generated error in a client's after-visit summary or portal notes yet?