In a new piece in Fierce Healthcare, Peter Grantcharov, CTO of Aimbient, and David Knobel, Commercial Counsel of Aimbient, tackle a problem they’ve seen growing in hospital AI committees: one-size-fits-all review that can under-govern high-risk generative AI models while blocking narrow models designed to reduce privacy exposure.

Most AI review committees ask one question: “Was PHI used to train this model?” But it’s the wrong one — it can allow riskier models to pass through while blocking narrow ones that don’t pose the same exposure risk. The piece lays out five better questions to ask instead.

At Aimbient, we know the surgeons and perioperative leaders driving this work inside their institutions are often the ones left waiting on a review process that isn’t asking the right questions. As partners, we wanted to give them something to point to.