From News to Intelligence

Anticipating developments that may affect migration requires timely information. Official statistics provide an essential evidence base, but they often became available after a significant delay. News media offer a faster, complementary signal of the events that can influence migration flows. 

The JRC report  explores how this potential can be harnessed through a prototype dashboard monitoring migration-related news from seven countries of origin, transit and departure on routes to the EU: Algeria, Egypt, Libya, Mali, Morocco, Niger and Tunisia. The work builds on the Europe Media Monitor (EMM), which processes around 500,000 news articles every day from approximately 20,000 sources worldwide. 

Between April 2024 and September 2025, the team identified more than 24.000 migration-related news items in Arabic, English and French. A large language model was used to propose topics for each article, while human experts reviewed and validated a sample of more than 2.000 news items. This process resulted in a taxonomy of 18 themes, covering situations ranging from irregular migration and smuggling to diplomacy, conflict and environmental factors that can influence migration. The validated dataset was used to train a multilingual classifier capable of assigning these themes automatically. 

The resulting demonstrator allows analysts to explore news by country, language, week and theme, to identify trends, and to request AI-generated summaries of selected articles. It can also help connect peaks in media attention to specific developments, such as political transitions, legislative changes or security incidents, which may act as precursors or triggers of changes in mobility. 

The work supports the Situational Awareness and Early Warning Framework (SAEWF) within the Migration Preparedness and Crisis Blueprint Network, overseen by the Directorate-General for Migration and Home Affairs. It was refined through workshops with policy officers while human experts remained central to defining, reviewing and validating the thematic framework. 

Next steps include integrating the classifier into the Commission's situational awareness toolbox, extending coverage to more countries, and exploring automated extraction of migration-related events. 

Beyond this specific demonstrator, the report provides a detailed account of the methodology and lessons learned, offering a practical starting point for others exploring how news monitoring and AI can be combined to strengthen situational awareness and early warning.