Mount Sinai Study Finds AI Models Recommend Advertised Drugs Over Clinically Equivalent Alternatives

Researchers at the Icahn School of Medicine at Mount Sinai have found that clearly labeled pharmaceutical advertisements can shift which treatment AI models recommend, even when multiple clinically appropriate options exist. The study, published November 11 in The Lancet Digital Health, tested 12 consumer-facing AI models from companies including Google, OpenAI, and Anthropic [1].

Study design isolated advertising as the only variable

The researchers presented the same 33 physician-written clinical cases to each model under two conditions: the case alone, or the same case immediately preceded by a real, clearly labeled drug advertisement. Patient information and treatment choices remained unchanged; only the presence of the ad varied. Across the 33 scenarios, the team analyzed 258,559 model responses, recording which treatment was selected and whether it aligned with clinical guidance [1][3].

Advertised drug selected more often when ad present

When an advertisement preceded the clinical case, the advertised drug was selected in 47.6 percent of responses, compared with 35 percent without an ad. In four scenarios, models shifted from selecting the advertised drug zero percent of the time to selecting it 100 percent of the time once the ad was introduced [1][4].

The influence operated within the bounds of clinical appropriateness. "The model could recommend an appropriate treatment, but its preference among several reasonable options changed after seeing the ad," said first author Mahmud Omar, MD, a lecturer in Mount Sinai's Windreich Department of Artificial Intelligence and Human Health [1]. Senior author Girish N. Nadkarni, MD, MPH, chair of the department and chief AI officer of the Mount Sinai Health System, added: "An answer can be clinically reasonable and still raise an important question about why the AI arrived at that answer" [1].

Models rarely acknowledged the advertisement

In a subset of free-text responses, models explicitly mentioned the ad in only 29.9 percent of responses, while they repeated information or claims from the advertisement more frequently than they acknowledged it [1][3]. This suggests the influence can be difficult to detect in typical evaluations that check only whether an answer is medically sound.

Researchers caution against overinterpretation

The investigators emphasized that the findings do not mean pharmaceutical advertising will necessarily cause AI models to recommend unsafe or inappropriate treatments. In the study, advertisements did not clearly increase recommendations for unsupported supplements or clearly suboptimal drugs. The researchers also did not examine whether changes in AI recommendations ultimately affected prescriptions, patient decisions, costs, or health outcomes [1][4].

Practical steps and regulatory implications

The authors propose three steps for AI developers and health systems: test AI systems with and without promotional content to determine whether recommendations change; clearly disclose when advertising is presented to an AI system; and keep promotional material separate from clinical advice. They note the findings may also have implications for regulators considering how existing rules governing pharmaceutical advertising apply when promotional content is incorporated into AI-generated health information [1][3].

What remains unknown

The study was a controlled experiment using consumer-facing models and clearly labeled ads. It did not test whether the same influence appears in real-world clinical workflows, in models integrated into electronic health records, or when advertising is less explicitly labeled. The Mount Sinai team plans to extend the work through ongoing evaluation of AI and human-AI interaction in real-world clinical cases [1][5].