Drug Ads Influence AI Treatment Recommendations, Mount Sinai Study Reveals 12.6% Shift

Pharmaceutical Advertisements Influence AI Treatment Recommendations

A groundbreaking study published in The Lancet Digital Health1

1

reveals that drug ads can significantly influence AI treatment recommendations, raising concerns about commercial bias in AI as people increasingly rely on artificial intelligence for health information. Researchers at the Icahn School of Medicine at Mount Sinai2

2

tested 12 AI models from major technology companies including Google, OpenAI, and Anthropic to determine whether clearly labeled pharmaceutical advertisements could shift which treatments AI systems recommend.

Study Reveals 12.6 Percentage Point Shift in Drug Selection

The research analyzed 258,559 AI responses across 33 clinical scenarios1

1

and found that the advertised drug was selected in 47.6% of responses when an ad was present, compared with 35% without an ad. This represents a 12.6 percentage point increase in recommendation rates for advertised medications. In four cases, models shifted from selecting the advertised drug 0% of the time to selecting it 100% of the time when the ad was introduced2

2

, demonstrating the potential magnitude of commercial influence on AI models.

Subtle Commercial Influence on AI-Generated Health Advice

First author Mahmud Omar, MD, a lecturer in the Windreich Department of Artificial Intelligence and Human Health, highlighted a particularly concerning aspect of the findings. "The model could recommend an appropriate treatment, but its preference among several reasonable options changed after seeing the ad,"1

1

Omar explained. Because the final answer could still appear medically sound, the influence of pharmaceutical advertising and drug recommendations could easily be missed in typical research evaluations and patient-facing use cases.

AI Models Rarely Acknowledge Promotional Content

The study examined how often AI models acknowledged the presence of advertisements or repeated information from them. In a subset of free-text responses, models explicitly mentioned the ad in only 29.9% of responses2

2

, while they repeated information or claims contained in the advertisement more frequently. This lack of transparency raises questions about whether users can effectively distinguish clinical guidance from commercial influence when interacting with AI systems.

Researchers Isolate Effect of Pharmaceutical Advertisements

The study was designed to isolate the potential effect of advertising on AI treatment recommendations. Researchers gave the AI models identical clinical cases, either alone or with a clearly labeled real drug advertisement placed immediately before the case1

1

. The patient information and treatment choices remained unchanged, with only the presence of the advertisement varying. This controlled approach allowed researchers to attribute differences in recommendations directly to the influence of promotional content.

Implications for AI Developers and Health Systems

Senior author Girish N. Nadkarni, MD, MPH, Chair of the Windreich Department of Artificial Intelligence and Human Health and Chief AI Officer of the Mount Sinai Health System, emphasized the need for transparency. "As AI becomes part of how people get health information, we need to be able to distinguish clinical guidance from commercial influence and be transparent about when advertising is part of the interaction,"2

2

Nadkarni stated. The findings point to several practical steps for AI developers and health systems, including testing AI systems with and without promotional content to determine whether recommendations change, clearly disclosing when advertising is presented to an AI system, and keeping promotional material separate from clinical advice.

Regulatory Discussions and Future Research

The findings may have implications for regulators as they consider how existing rules governing pharmaceutical advertising apply when promotional content is incorporated into AI-generated health advice1

1

. The researchers noted that advertisements did not clearly increase recommendations for unsupported supplements or clearly suboptimal drugs, and they did not examine whether changes in AI recommendations ultimately affected prescriptions, patient decisions, costs, or health outcomes. The research team plans to extend the work through their ongoing evaluation of AI and human-AI interaction in real-world clinical scenarios2

2

, which will provide crucial insights into the long-term effects of commercial bias in AI on healthcare delivery.