Drug or Pokémon? An analysis of the ability of large language models to discern fabricated medications
LLMs are susceptible to adversarial attacks, especially with medications. Further model improvement is imperative before LLMs are considered safe and reliable for routine use in the medical field.
Authors
- Susan Smith (ORCID: https://orcid.org/0000-0002-5171-8405)
- Seth R. Bauer (ORCID: https://orcid.org/0000-0002-0420-0320)
- Sunghwan Sohn (ORCID: https://orcid.org/0000-0001-8256-2602)
- Xingmeng Zhao (ORCID: https://orcid.org/0000-0002-7346-758X)
- Tellen D. Bennett (ORCID: https://orcid.org/0000-0003-1483-4236)
- Kaitlin Blotske (ORCID: https://orcid.org/0000-0002-2977-0266)
- Raja-Elie E Abdulnour
- Tianming Liu
- Mitch Cohen
- Brian Murray
- Yanjun Gao
- Brooke Smith
- Kelli Henry
- Erin Barreto
Institutions
- Brigham and Women's Hospital (US)
- WellStar Health System (US)
- Cleveland Clinic (US)
- Harvard University (US)
- University of Georgia (US)
- Augusta University (US)
- Augusta University Health (US)
- Mayo Clinic in Arizona (US)
- University of Colorado Anschutz Medical Campus (US)
- University of Montana (US)
- University of Colorado Denver (US)
Publication Details
- Journal
- medRxiv
- Published
- 2026-01-13
- DOI
- https://doi.org/10.64898/2026.01.12.26343930
- Primary Topic
- Artificial Intelligence in Healthcare and Education
- Type
- article
- Field-Weighted Citation Impact
- 0.00