Drug ads shifted AI models’ treatment recommendations in a study

Researchers at the Icahn School of Medicine at Mount Sinai tested 12 AI models from major technology companies using physician-written clinical cases.

They compared responses when each case appeared alone with responses when a clearly labeled drug ad came first, across 33 scenarios.

Models chose the advertised drug in 47.6% of responses with an ad, compared with 35% without one.

In some answers, models repeated claims from ads without acknowledging the ad, even when their recommendations still seemed medically reasonable.

The study did not show that ads led to unsafe treatment choices, and it did not assess effects on prescriptions, patient decisions, costs or health outcomes.

The researchers say developers and health systems should test whether promotional content changes AI recommendations and clearly disclose when ads are shown.