EMA has taken steps to explore the use of AI in regulatory activities, including:
EMA collaborates closely with the Heads of Medicines Agency (EMA) on using and integrating AI tools.
For more information on EMA's work on AI, see:
EMA identified the main uses cases for AI within the network. They are:
EMA collected AI use cases and explored solutions to implement them via a series of workshops with national competent authorities (NCAs).
These workshops covered both human and veterinary domains. They helped identify more than 60 AI uses cases.
An AI tools framework maintains a catalogue of tools that support the main AI use cases within the network. These tools use technologies including:
The framework enables the network to use AI solutions more efficiently, maximising value and regulatory impact.
It creates a common way for EMA, NCAs and the European Commission to share and access information about AI tools.
The joint EMA-HMA Network Data Steering Group (NDSG) adopted this framework in 2025.
EMA is helping develop an ecosystem of AI agents and assistants to optimise and enhance knowledge and information retrieval.
As part of its broader work to develop and deploy AI solutions, EMA has also tested a prompt community model for the network. This aimed to create a collection of validated and reusable prompts.
The prompt community supported:
The NDSG established this pilot in 2025 and tested it in 2026.
For more information, see:
EMA extended the functionality of its AI-enabled knowledge mining tool called 'Scientific Explorer' to support EMA and national competent authorities (NCAs) in finding information related to initial marketing authorisation applications of human medicines. This includes public assessment reports.
The extended search functionality is available as of March 2026.
The tool enables easy, focused and precise search of regulatory scientific information from network sources.
EMA and NCA assessors are able to use the tool to customise their search for related information based on medicine name, disease or relevant keywords.
This supports decision-making via efficient retrieval of regulatory precedents.
EMA introduced the tool in March 2024, initially focusing on improving searches for EMA scientific advice procedures for human and veterinary medicines.
The tool is available only to approved and authenticated users representing the network.
For more information, see the document below with answers to frequently asked questions about this tool:
EMA has qualified the use of AIM-NASH, an AI-assisted tool that helps pathologists assess liver biopsy samples in clinical trials for metabolic dysfunction-associated steatohepatitis (MASH). This condition was formerly known as non-alcoholic steatohepatitis (NASH).
The tool uses a machine learning model trained on thousands of liver biopsy samples and pathology annotations. It is intended to support expert assessment. A qualified pathologist must review the AI-generated results and remains responsible for the final evaluation of each biopsy sample.
The tool helps researchers obtain clearer evidence on the benefits of treatments in clinical trials with fewer patients.
PathAI designed this tool to assist, but not replace, expert human review.
AIM-NASH is the first AI-based methodology for medicine development to receive a qualification opinion from EMA's human medicines committee (CHMP).
CHMP concluded that AIM-NASH can help reduce variability in biopsy assessments, and improve the consistency and reproducibility of results in MASH clinical trials.
EMA's human medicines committee based its conclusion on evidence showing that AI-assisted assessments, verified by a pathologist, can reliably determine disease activity and fibrosis stage.
Marketing authorisation applicants may continue to submit clinical trial data generated using AIM-NASH, if the tool is used in line with the qualified conditions set by CHMP.
The qualification of AIM-NASH also marks an important step in the use of AI in medicine development. It demonstrates that AI-based methodologies can be assessed through EMA's established qualification framework.
EMA issues this qualification opinion in March 2025.
For more information on this tool, and on EMA's first qualification opinion on AI methodology, see the documents included below:
EMA uses two versions of Microsoft's Copilot AI tool to help streamline the work and boost the productivity of its staff:
While Copilot 365 is widely used at EMA, it is not available to the wider network. It is also not approved for use in cross-network medicine regulation processes.
EMA has put in place enteprise-grade security features before adopting Copilot. These features ensure that EMA's organisational data remain protected at all times. They include:
In addition, EMA regularly reviews the use of AI and associated risks as part of its AI risk-management arrangements.
For more information on use of AI at EMA, see:
Visitors of this website can use a translation feature to better understand EMA’s role and work in their preferred official EU language.
The European Commission developed this feature in line with standards for EU institutions.
This feature uses artificial intelligence, more specifically neural machine translation, to automatically translate texts into other languages. It is not a generative-AI chatbot, but a dedicated machine translation tool built on the EU institutions’ decades of professional translation expertise.
EMA launched this feature on this website in June 2026.
For more information on EMA's machine translation feature, see:
An update log is available to show the date and summary of changes to this webpage. It does not include updates to linked documents or minor edits like typos or broken link fixes.
The tracking of updates begins in November 2026.
11 November
Page first published, containing information transferred from the main AI page, such as the 3 March 2026 update on the extended functionality of EMA's AI-enabled knowledge mining tool Scientific Explorer.