Artificial intelligence (AI) allows computers and machines to simulate learning, comprehending, problem-solving, and decision-making in a similar manner to humans.1 AI is rapidly transforming the field of pharmacy and will continue to do so.
The benefits of AI in pharmacy are far-reaching. It can enhance clinical and personalized medicine through pharmacogenomics, clinical decision support systems, and predictive analytics for medication management. It can streamline pharmacy operations through workflow automation, inventory optimization, billing, and regulatory compliance. AI improves patient safety and engagement by way of adverse event monitoring, patient engagement tools, and educational resources. It also accelerates drug discovery and development through clinical trial optimization and drug repurposing.
Although these benefits are crucial to the field of pharmacy, integration of AI presents significant ethical challenges that must be recognized and addressed to ensure patient welfare. Careful consideration must be given to key ethical concerns such as data privacy, algorithmic bias, transparency, accountability, and human oversight in order to balance crucial innovation with ethical standards.
Data Privacy and Security
AI relies on vast amounts of sensitive patient data, thereby creating significant data privacy challenges. Proper safeguards are necessary to protect patient data and adhere to relevant regulatory requirements, such as the Health Insurance Portability and Accountability Act (HIPAA).2
Data Breaches and Cyber-Attacks
AI-driven pharmacies are vulnerable to data breaches and cyber-attacks through the targeting of patient information, intellectual property, and drug data. An increased reliance on digital systems creates entry points for attackers to use ransomware, commence phishing, and breach AI data pipelines.3
Re-Identification of De-Identified Data
Combining datasets using advanced algorithms to exploit unique combinations of identifiers that can then be used to find overlapping information. These can then be pieced together to uncover a real identity.4
Third-Party Vendor Risk
In health care, almost one-third of all data breaches originate through third-party compromise. These attacks target vulnerabilities in the vendor’s systems, allowing attackers to access data they normally would not have access to.5
Algorithmic Bias and Fairness
Most AI algorithms learn from enormous datasets. If these datasets are incomplete, skewed, or compromised in any way, the AI algorithm may reproduce and amplify incorrect information.
Biased Training Data
If training data is misrepresentative of the full population, AI may propagate this bias, which may lead to poor patient outcomes.6
Lack of Data Diversity
Data diversity represents the inclusion of data from a variety of demographics including age, gender, ethnicity, geographic locations, and socioeconomic status. Cultural diversity indicates accounting for differences in language, customs, and mores. Behavioral diversity denotes including data from more than just dominant patterns. Finally, contextual diversity means that data is captured from varied environments and situations.6
Information Bias
Information bias is the outcome of a combination of errors in how training data is collected, measured, or processed that can lead to inaccurate and possibly discriminatory results.
Unintended Feedback Bias
AI models learn not only through large datasets but also through a feedback loop. This can lead AI to reinforce its own flawed conclusions.
Transparency and Explainability
By making AI processes and data understandable and providing clear and understandable reasoning for recommendations produced by AI, pharmacists can help detect bias, ensure accuracy, and enhance patient safety. This also facilitates shared decision-making by helping both patients and health care providers understand the risks and benefits of therapy.
Human Oversight and Control
Incorporating human intervention throughout the AI process is imperative to ensure patient safety, ethical practice, regulatory compliance, and system accuracy. AI cannot match pharmacists’ strategic thinking, ethical judgment, or clinical expertise. Regulators of AI must focus on human-in-the-loop (intervention at critical points to guide and review AI outputs) and human-on-the-loop (human supervision for necessary interventions).7
Equitable Access
Health equity relating to AI applications refers to the fair and just distribution of health technologies and their associated benefits. All individuals should have access to the same health care services regardless of race, gender, ethnicity, socioeconomic status, or geographic location.
Artificial intelligence must advance health care and provide positive health outcomes while also doing so in a way that lessens existing health disparities rather than increasing them.8
Accountability and Liability
Accountability and liability in pharmacy when using AI is complex as there are several stakeholders to consider and a lack of clear legal frameworks.
- Pharmacists: Pharmacists are responsible for patient care and are required to act under the “reasonable professional” standard whether they are using AI as a tool. Failure to act on a flawed AI recommendation can result in liability and, conversely, failure to use an available AI tool that may have prevented an error are liable as well.
- Health care institutions: Hospitals and pharmacies can be held liable for the AI they implement. Negligent supervision encompasses failure to adequately train staff and set oversight procedures. Negligent selections would be implementing an unreliable and/or unverified AI system. Vicarious liability includes holding the institution responsible for the actions of their employees.
- AI developers and vendors: Product liabilities include coding errors and glitches, inadequate prerelease testing, failure to warn of limitations and risks of products, and using biased training data that leads to discriminatory outcomes.
Patient Autonomy and Informed Consent
Patients must understand how AI can influence treatment decisions through clear communication about the process in which AI analyzes data and makes recommendations. Patients have the absolute right to make informed decisions about their care, even when AI is involved.
Regular Monitoring and Auditing
To remain reliable and effective, AI systems must be continuously monitored and evaluated for “drift”, where the accuracy of AI recommendations degrades over time due to data changes or evolving circumstances.
Conclusion
The potential for AI to transform the field of pharmacy is immense. This potential is tempered by a wide range of ethical considerations including data privacy, bias, transparency, equitable access, and accountability. Successful implementation of AI requires collaborative efforts between AI developers, pharmacists, regulators, and more to establish ethical standards and educational frameworks.
REFERENCES
-
What is AI? IBM website. August 9, 2024.
https://www.ibm.com/think/topics/artificial-intelligence . Accessed September 9, 2025. -
Artificial Intelligence in Pharmacy: Appropriate Use of AI (Part 1 of 2). Florida Pharmacy Foundation website.
https://flpharmfound.org/artificial-intelligence-in-pharmacy-part-1#:~:text=Additionally%2C%20the%20ethical%20implications%20of,and%20mitigate%20potential%20ethical%20issues . Accessed September 9, 2025. -
Li J. Security Implications of AI Chatbots in Health Care. J Med Internet Res. 2023 Nov 28;25:e47551. doi: 10.2196/47551. PMID: 38015597; PMCID: PMC10716748.
-
Erosion of Anonymity: Mitigating the Risk of Re-Identification of De-Identified Health Data. Health Law Advisor website. February 28, 2019.
https://www.healthlawadvisor.com/erosion-of-anonymity-mitigating-the-risk-of-re-identification-of-de-identified-health-data#:~:text=To%20mitigate%20these%20risks%2C%20organizations%20can:%20*,sharing%20and%20use%20agreement%20with%20the%20recipient . Accessed September 10, 2025. -
More Than One-Third of Data Breaches Due to Third-Party Supplier Compromises. The HIPAA Journal website. March 28, 2025.
https://www.hipaajournal.com/more-than-one-third-data-breaches-third-party-compromises/ . Accessed September 10, 2025. -
Norori N, Hu Q, Aellen FM, Faraci FD, Tzovara A. Addressing bias in big data and AI for health care: A call for open science. Patterns (N Y). 2021 Oct 8;2(10):100347. doi: 10.1016/j.patter.2021.100347. PMID: 34693373; PMCID: PMC8515002.
-
Singh R, Paxton M, Auclair J. Regulating the AI-enabled ecosystem for human therapeutics. Commun Med (Lond). 2025 May 17;5(1):181. doi: 10.1038/s43856-025-00910-x. PMID: 40382515; PMCID: PMC12085592.
-
Health Equity and Ethical Considerations in Using Artificial Intelligence in Public Health and Medicine. Centers for Disease Control and Prevention website. August 22, 2024.
https://www.cdc.gov/pcd/issues/2024/24_0245.htm#:~:text=The%20potential%20of%20AI%20to,or%20geographic%20location%20(8) . Accessed September 10, 2025.