AI@EFSA

AI@EFSA

Artificial Intelligence (AI) and, most recently, the arrival of Generative AI (GenAI), mark a transformative shift and promise exponential increases in productivity and efficiency in the private and public sectors. As stated by the European Commission (EC) in its Communication Artificial Intelligence in the European Commission (AI@EC) (European Commission, 2024), AI brings "unprecedented possibilities for supporting our staff, reducing administrative burden, and enhancing quality and impact of the Commission's work".

EFSA began exploring AI in 2017 and investigated its potential in risk assessment (RA). Recognising the need for a broader strategy, EFSA published its thematic paper on Artificial Intelligence in risk assessment (Cavalli et al., 2022), followed by the Roadmap for actions on AI for evidence management in risk assessment (Bersani et al., 2022). These initiatives laid the foundation for human-centric AI adoption to improve evidence management by 2027.

In January 2024, EFSA established an AI Task Force to identify incremental actions that can be undertaken with a view to exploiting the power of emerging AI technologies. The EFSA AI Task Force, in line with the strategic vision outlined in AI@EC (European Commission, 2024), worked on three tasks:

• supporting the development of an internal network of interested colleagues;
• identifying AI-enabled information technology (IT) investments needed for AI adoption;
• supporting the definition of internal guidelines in line with recommendations and guidelines from relevant EU audit and control bodies, including the European Ombudsman and the European Data Protection Supervisor (EDPS) and/or any guidance provided by the EC.

In particular, EFSA aims at setting clear goals and principles for the adoption of human-centric AI in its main processes. This effort requires robust AI governance to ensure compliance and accountability - prerequisites for using AI in EFSA's core processes and to assist in the analysis of evidence submitted by applicants or collected through literature reviews, outsourcing procedures, or internal efforts.

The AI Task Force terminated its mandate in December 2024 and handed over its activities to the newly established governance structures introduced in this document.

Documents