“This request for information provides an opportunity for interested parties and the public to share their input” Food and Drug Administration

The proposed programme, titled AI-Enabled Optimization of Early-Phase Clinical Trials, aims to explore whether AI can improve some of the most challenging aspects of drug development, including patient recruitment, dose optimisation, safety monitoring, adaptive trial design and biomarker-driven patient selection.

The most important aspect of the announcement may not be the tech or framework itself, but the fact that the regulators are actively seeking feedback before establishing expectations. Stakeholders are being asked how AI performance should be measured, how fairness should be assessed across different patient populations, how model drift should be monitored and what is the bare minimum level of explainability we should expect when AI influences clinical decisions. In short, the FDA is inviting the industry to help draft the playbook before it is written.

For players in the pharma industry, this consultation gives a chance to influence future requirements around validation, governance and oversight so it is based on real world experience and ensures expectations are firm but grounded and feasible. The standards emerging from this pilot could eventually shape how regulators evaluate AI across the wider clinical development landscape.

Perhaps the strongest message from the FDA is that these expectations are not yet fixed. The agency specifically wants organisations of different sizes and levels of AI adoption to contribute, acknowledging the requirements must be practical and equitable for smaller firms.

The consultation window closes on 29 June, making this an incredible opportunity for pharmacovigilance organisations to influence how AI regulation develops in clinical research.

So, if your PV firm works under American regulation and perhaps want to give input into what can be realistically expected from smaller organisations or feel strongly about a way in which these systems should be validated, monitored and governed, now is the time to speak. Those who speak will help to shape the future of AI drug safety regulation; those who do not may find themselves working within a ruleset designed by their competitors.