The European Medicines Agency (EMA) and the US Food and Drug Administration (FDA) have jointly established new AI principles in drug development to reduce regulatory divergence between the major markets of the European Union and the United States.

Industry associations have applauded the landmark accord, as it strengthens harmonisation across the regions – but emphasise that more concrete steps are needed.

With AI technologies becoming increasingly embedded in evidence generation or analysis in drug development, regulators are pivoting from monitoring to establishing principles-based guardrails to improve the accountability, integrity, and performance of the new technology.

The accord is likely to have a significant effect on global AI use in drug development, as the regulatory weight of the EMA and the FDA decisions sets global standards.

AI principles are ‘first step in renewed cooperation’

The AI principles, published on Wednesday, (14 January), mark the culmination of a process that began two years ago to address the regulatory divergence between the major markets EU and the US. This was proving to be a significant barrier to digital innovation in pharma.

“The guiding principles of good AI practice in drug development are a first step of a renewed EU-US cooperation in the field of novel medical technologies,” said Olivér Várhelyi, European Commissioner for Health and Animal Welfare,

The goal is to preserve a “leading role in the global innovation race, while ensuring the highest level of patient safety,” he emphasised.

Industry associations welcome the accord

“We view the joint work of the EMA and FDA on developing these principles as a positive and important step toward global regulatory convergence,” the European Federation of Pharmaceutical Industries and Associations (EFPIA) told Health Policy Watch.

With memberships of 36 national associations and 40 pharmaceutical companies, EFPIA represents the business interests of the biopharmaceutical industry operating in Europe.

A press query with the Pharmaceutical Research and Manufacturers of America remained unanswered.

The AI principles aim to govern the use of the new technology across its entire lifecycle in drug development – from early-stage drug discovery and clinical trials to manufacturing and post-market safety monitoring.

This is to ensure that patient safety and ethical integrity remain paramount, emphasising a human-centric approach and oversight, as the US and EU regulatory bodies state.

With the new principles, the EMA and the FDA aim to dismantle the “AI black box”, as industry experts call it, with users and patients often failing to understand the process behind the generation of results.

According to the authorities, AI must be applied in well-defined contexts and be understandable: pharmaceutical or biotechnology companies are expected to use plain language to explain AI limitations and underlying data, while adhering to the protection of privacy and sensitive information.

AI principles to enhance accountability

Crucially, the joint document addresses a major issue in pharmaceutical operations known by industry experts as “shadow use”, namely that analysts are already working with large language models (LLMs) for daily tasks while leadership regularly looks the other way. This is to be countered by better oversight.

The provisions also aim to ensure clinical teams understand the technical tools being used by analysts by mandating that data scientists are integrated with clinical leads throughout the lifecycle of drug development.

Moreover, instead of single validation, continuous monitoring for “data drift” is required – a phenomenon where an AI’s performance degrades over time as the underlying data environment changes.

High-level declaration needs further steps

As the new AI principles are strategically designed to align with existing regulations, the guiding principles do not fundamentally change how AI is already being used by European industry, EFPIA explains.

However, they create the basis for a unified language in medical technology development, possibly with wide-ranging implications for global AI regulation in drug development.

“They help create a more coherent environment for scaling AI tools globally and for engaging with regulators in a consistent manner,” said the EFPIA.

While the guidelines mark a good foundation for the industry to lower the risk of duplicative or divergent requirements across regions, the AI principles remain high-level. Further steps toward concrete shared terminology, definitions, and concepts across regions are needed. Some of which are to be expected this year.

Image Credits: European Union/Christophe Licoppe, European Union/Frédéric Garrido-Ramirez, Felix Sassmannshausen, European Union/Barcelona Supercomputing Center , EMA.