Recent regulatory developments are shifting the landscape for AI in drug development, as organizations pivot from experimentation to establishing robust AI systems that meet regulatory standards.
On January 14, the European Medicines Agency (EMA) and the US Food and Drug Administration (FDA) released joint guiding principles aimed at defining good AI practices in drug development. These principles serve as a reference for integrating AI into regulated processes, emphasizing the need for risk-based controls, data governance, and lifecycle management. This marks a transition from merely assessing AI's feasibility to ensuring its reliability, transparency, and accountability in influencing critical decisions.
Central to these guidelines is the concept of 'context of use,' which highlights that AI should be evaluated based on its intended application and associated risks. For instance, the use of AI for summarizing internal documents differs significantly from its application in safety reviews or trial operations, which demands stricter controls and oversight. As AI becomes more embedded in regulated workflows, organizations must carefully map use cases to their risk profiles, ensuring clear accountability and documentation.
The principles also stress that lifecycle management is crucial when AI is integrated into regulated environments. Continuous monitoring and version control of AI outputs are essential to maintain compliance and accountability. As regulatory frameworks evolve, particularly with the upcoming EU AI Act and FDA guidance on electronic systems, organizations are encouraged to operationalize these principles rather than slow down AI adoption. This proactive approach will help integrate AI into existing governance structures, ultimately enhancing the integrity and reliability of evidence generation in drug development.