EMA and FDA have set out ten common principles for the use of artificial intelligence in drug development, establishing a basis for future regulatory guidance in Europe and the United States.

Artificial intelligence (AI) is gaining relevance across different stages of drug development, from early research to post-marketing safety monitoring. To keep pace with this evolution, regulatory authorities are working on principles and guidance that allow the potential of these technologies to be harnessed without compromising the quality, safety and efficacy of medicines.

In this context, the European Medicines Agency (EMA) and the United States Food and Drug Administration (FDA) have established ten common good practice principles for the use of AI in drug development. The initiative, announced in January 2026, represents an important step in international cooperation and in building a more consistent regulatory approach to these technologies.

Ten principles for the responsible use of AI

The principles cover the use of AI in generating and analyzing evidence across the medicines life cycle, including the nonclinical, clinical, manufacturing and post-marketing phases.

The document identifies ten key areas:

1. Human-centric by design: The development and use of AI technologies are aligned with ethical and human-centric values.
2. Risk-based approach: The development and use of AI technologies follow a risk-based approach, with proportionate validation, risk mitigation measures and oversight appropriate to the context of use and to the risk determined for the model.
3. Adherence to standards: AI technologies comply with relevant legal, ethical, technical, scientific, cybersecurity and regulatory standards, including Good Practices (GxP).
4. Clear context of use: AI technologies must have a clearly defined context of use, including the role they play and the purpose for which the system will be used.
5. Multidisciplinary expertise: Multidisciplinary expertise covering both the AI technology and its context of use, including appropriate scientific, technical and regulatory knowledge, is integrated throughout the technology life cycle.
6. Data governance and documentation: The provenance of data sources, processing steps and analytical decisions is documented in a detailed, traceable and verifiable manner, in line with GxP requirements. Governance of the data used must ensure its appropriate management, including privacy and the protection of sensitive data, throughout the technology life cycle.
7. Model design and development practices: The development of AI technologies follows good design and engineering practices, using data fit for the intended purpose and promoting interpretability, explainability, reliability, robustness and patient safety.
8. Risk-based performance assessment: Performance assessment of AI must be risk-based, cover the complete system, including interaction with users, and rely on data, metrics and validation methods appropriate to the intended context of use.
9. Life cycle management: Risk-based quality management must cover the entire AI life cycle, including the identification and resolution of issues, scheduled monitoring and periodic re-evaluation of performance. AI systems are monitored during use, including the management of relevant changes.
10. Clear, essential information: Information on AI technologies must be clear, accessible and appropriate to the target audience, covering their purpose, performance, limitations, data used, updates and interpretability.

Taken together, these principles seek to establish a common basis for assessing and managing AI applications, without replacing the regulatory requirements applicable to the quality, safety and efficacy of medicines.

What could change in the regulatory framework?

The initiative establishes a common basis for the development of future, specific guidance on the use of AI in drug development and strengthens cooperation between EMA and FDA.

In the European Union, this work builds on the reflection started by EMA in 2024, through the publication of the document “Reflection paper on the use of Artificial Intelligence (AI) in the medicinal product lifecycle”, which addresses the use of AI throughout the life cycle of medicines. See below the original full post published on Vita’s LinkedIn in July 2023:

Additionally, in May 2025, EMA and the EU national agencies, brought together in the Heads of Medicines Agencies (HMA) group, also launched the strategic plan “Data and AI in Medicines Regulation to 2028”, which provides guidance on the use of data and AI in medicines regulation. In July of the same year, the European authority published the first report of its AI Observatory, which compiles the experience of the European regulatory network and identifies applications of AI across the entire medicinal product life cycle. These initiatives show that European regulation has been preparing, progressively, for the use of AI (see the full post on the Vita blog).

Future European guidance is expected to complement the ten common principles, taking into account the applicable legal requirements and the evolving use of AI in the pharmaceutical sector, with a view to promoting greater clarity and consistency in regulatory expectations.

For pharmaceutical companies, it will be important to follow these developments and assess the AI applications used or planned in their processes, particularly with regard to data quality, documentation, performance validation and risk management.

Cooperation between EMA and FDA may also contribute to greater convergence of regulatory approaches. However, the existence of common principles does not mean that the legal requirements or procedures of the two authorities will necessarily be identical.

It should be noted that the publication of the ten principles does not, in itself, create new uniform legal obligations. Their practical application will depend on the regulatory requirements and relevant guidance in each jurisdiction.

Evolution of AI applications in the pharmaceutical sector

AI may help speed up certain research and development activities, support data analysis, strengthen pharmacovigilance and improve the identification of potential safety signals.

Its use in activities related to regulatory assessment is also being considered in the context of the evolution of European pharmaceutical legislation.

These benefits are potential and depend on the suitability, reliability and validation of the applications used. Adopting the technology does not remove the need for sufficient scientific evidence, nor the responsibility of the parties involved.

What should pharmaceutical companies follow?

For companies involved in the development and manufacture of medicines, as well as for applicants and marketing authorization holders (MAHs), it will be important to follow the new guidance and assess the AI applications already used or planned in their processes.

In the European Union, it will be particularly important to follow future guidance from EMA and the European Medicines Regulatory Network, taking into account the applicable legislation and the evolution of the European framework on AI.

For companies operating in Brazil, Europe and/or the United States, it will be equally important to distinguish the common international principles from the requirements specific to each jurisdiction, following the guidance of the competent authorities in each market.

The use of AI in the Brazilian context has also been signaled by ANVISA, when it used this tool for the analysis of impurities (see the full post on the Vita blog).

As the technology evolves, following regulatory developments will be essential for pharmaceutical companies to assess opportunities, anticipate the need for adaptation and ensure that the use of AI remains compatible with the quality, safety and efficacy requirements for medicines.