The European Medicines Agency (EMA) has published its 2025 AI Observatory report, assessing the use of artificial intelligence across the medicines lifecycle. The document, released on 4 June 2026, outlines the agency's findings on AI adoption in drug development, regulatory decision-making, and pharmacovigilance, and identifies key opportunities and challenges for stakeholders including pharmaceutical companies, regulators, and patients.
The report is the latest in EMA's series monitoring AI integration in the pharmaceutical sector. It covers applications from drug discovery and clinical trials to manufacturing and post-market surveillance. The agency notes that AI is increasingly used to accelerate research, but also raises concerns about data quality, algorithmic transparency, and the need for robust validation frameworks.
Document type and status
The 2025 AI Observatory report is a non-binding strategic document that synthesises evidence from literature reviews, stakeholder consultations, and case studies. It does not set new regulatory requirements but aims to inform future policy and guidance. The report includes concrete examples of AI tools already in use, such as machine learning models for predicting drug toxicity and natural language processing for analysing adverse event reports.
Policy orientations and trade-offs
EMA identifies several policy priorities: promoting innovation while ensuring patient safety, developing clear standards for AI validation, and fostering international harmonisation. The report highlights a tension between encouraging rapid AI adoption and maintaining rigorous oversight. It calls for proportionate regulation that does not stifle innovation but ensures that AI-driven decisions are explainable and accountable.
Impact on stakeholders
Pharmaceutical companies may benefit from clearer expectations on AI use, potentially reducing regulatory uncertainty. However, they may face increased costs to comply with emerging validation and transparency standards. Regulators will need to invest in new expertise and tools to evaluate AI-based submissions. Patients could gain from faster drug development and more personalised treatments, but also face risks if AI systems are not adequately validated.
Expected institutional follow-up
EMA plans to use the report's findings to update its existing guidelines on AI in medicines, expected by late 2026. The agency will also continue its AI Observatory workstream, with a focus on developing practical tools for AI assessment and fostering multi-stakeholder dialogue. The report may inform broader EU digital health policy, including the European Health Data Space and the AI Act implementation.