Artificial Intelligence in Pharmaceuticals: Ethical Risks, Regulatory Challenges, and a Framework for Responsible Governance

Artificial intelligence (AI) is used in drug discovery, development, manufacturing, medication management, and patient care. These uses differ in supporting evidence and potential harm, so oversight must match their risks. This critical narrative review evaluates selected research, official US and Indian regulatory documents, and international and Indian ethical guidance. Concerns include biased or unrepresentative data, privacy, cybersecurity, unreliable outputs, limited transparency, divided responsibility, staff readiness, and unequal access. Much of the evidence comes from reviews, surveys, interviews, and limited performance tests. US Food and Drug Administration (US FDA) documents emphasize intended use, risk-based evidence of reliability, and lifecycle oversight. Discussion papers, draft guidance, and guiding principles differ from binding requirements. World Health Organization (WHO) and Indian Council of Medical Research (ICMR) guidance provide ethical foundations, while Indian medical-device rules and Central Drugs Standard Control Organization (CDSCO) software guidance connect intended use with regulation, local suitability, and monitoring. The review proposes seven governance stages: define intended use; assess risk and responsibilities; manage data and vendors; establish reliability; approve controlled introduction; monitor and manage changes; and address harm or end use. Each stage identifies responsibilities, evidence, and decisions. Assessment also considers human–AI performance, effects across groups, service resilience, and opportunities to question decisions. Responsible pharmaceutical AI requires continuing organizational oversight matched to effects on product quality, regulatory evidence, and patient care.