The European Medicines Agency (EMA) and the US Food and Drug Administration (FDA) have agreed on guiding principles for their cluster teleconferences on artificial intelligence (AI) in pharmacovigilance (PV), according to a document dated 2 March 2026 and published by EMA on 17 July 2026. The principles outline objectives, participants, timing, agenda setting, confidentiality, and record-keeping for the bilateral meetings, which aim to exchange information on AI-related policies, explore regulatory alignment, foster international collaboration, and share lessons learned on AI applications in PV.
The cluster will be co-chaired by EMA and FDA staff with expertise in AI and pharmacovigilance. Observers from Japan's Pharmaceuticals and Medical Devices Agency (PMDA) and Health Canada may participate, and observers from other regulatory authorities could join later subject to both agencies' agreement and appropriate confidentiality arrangements. Teleconferences are scheduled every two to three months, lasting 60 to 120 minutes, with agendas mutually agreed upon about two weeks in advance. Topics for discussion, listed in an appendix, include AI governance, validation of AI models, use of large language models, and real-world evidence generation.
Confidentiality is governed by existing bilateral information-sharing arrangements, and participants are prohibited from disclosing shared information without prior authorization. Short action points will be co-developed as summary records, and an online sharing platform may be used between meetings. The document, version 2 of EMA/422032/2023, updates earlier arrangements and reflects the agencies' commitment to structured collaboration on AI in pharmacovigilance, a field where both regulators face increasing pressure to integrate AI tools while ensuring patient safety and data privacy.
The guiding principles have moderate impact on stakeholders. For EMA and FDA, they formalize a framework for regulatory cooperation, potentially speeding up alignment on AI standards. For pharmaceutical companies, clearer regulatory expectations could reduce compliance uncertainty but may also impose additional documentation and validation requirements for AI-driven PV systems. Patients stand to benefit from improved safety monitoring through advanced AI tools, though concerns about data confidentiality and algorithmic bias remain. Observers from other regulators, such as PMDA and Health Canada, gain insight into EU-US approaches but face limited influence on agenda setting.