Health systems have rapidly deployed generative AI for clinical documentation, billing and patient communications, but industry leaders warn AI alone is insufficient for translating prescription labels patients rely on after leaving the pharmacy. Sharon Blank, CEO of RxTran and Language Scientific, argues the core problem is cultural: organizations treat language access as a compliance checkbox rather than a patient-safety imperative. Prescription labels require precise clinical nuance—dosage, timing, route, units and cautions—that machine translation and large language models can mishandle, increasing risk of misadministration, nonadherence and adverse events.
Experts recommend hybrid workflows that pair automated translation with domain-trained models, pharmaco-linguistic glossaries, medically qualified translators and pharmacist verification, plus continuous QA, auditing and governance. Prioritizing safety-oriented processes, measurement and human‑in‑the‑loop controls mitigates legal and equity risks and ensures multilingual labeling supports safe medication use rather than merely meeting regulatory obligations.
