The final edition of the STARLIGHT project annual newsletter offers an opportunity to reflect on a journey that began in October 2021 and has spanned 52 months of close collaboration between research, industry, and Law Enforcement Agencies (LEAs) across Europe. It is a chance to reflect not only on what STARLIGHT has delivered, but on how it has shaped collaboration, practice, and thinking around responsible AI for security.
The STARLIGHT Project Results brochure brings together the key outcomes and achievements of the project after more than four years of collaboration under Horizon 2020. It highlights STARLIGHT’s work on developing, testing, and assessing AI-based tools for civil security, with a strong focus on ethical, trustworthy, and responsible use in law enforcement contexts.
The STARLIGHT Final General Assembly took place on 14-15 January in Paris, hosted by the Institut des Hautes Etudes du Ministère de l'Intérieur (IHEMI). The two-day event marked the culmination of 52 months of collaboration dedicated to advancing responsible and trustworthy AI for European law enforcement.
The STARLIGHT project has successfully completed its 8th and final Operational Pilot, hosted by the Centre for Security Studies (KEMEA) in Athens on 25–26 November 2025. As the coordinating partner for all pilot activities, KEMEA led the organisation, testing, and validation processes, bringing together Law Enforcement Agencies (LEAs) and technical partners from the project’s 50+ member consortium.
On 17 October 2025, the Netherlands Forensic Institute (NFI) presented the STARLIGHT project at the ENFSI Forensic Information Technology Working Group (FITWG) Annual Meeting in Istanbul. The session, delivered on Day 3 of the event, highlighted STARLIGHT’s ethically aligned AI tools for law enforcement and forensics and drew strong interest from attending experts and practitioners.
The meeting brought together specialists from across Europe and beyond to exchange knowledge and explore advances in digital forensics and forensic IT.
STARLIGHT research was presented at the SPIE conference on Artificial Intelligence for Security and Defence Applications (15–18 September 2025, Madrid).
Sabina B. van Rooij (TNO) presented the paper, “Combining and enhancing global and local vision foundation models for explainable tattoo matching,” a collaboration with CERTH. The work demonstrates how combining global and local vision models can improve tattoo matching in criminal investigations while increasing explainability for end-users.
On 21st May 2025, the STARLIGHT project participated in the INVICTM Forum: From Crisis to Collaboration – Strengthening Victim Support through Cooperation, Technology, and Resilience, held in Lisbon. As one of the event sponsors, STARLIGHT supported this important platform for dialogue and exchange on advancing victim support through innovation and collaboration.
On 8th of April 2025, the STARLIGHT project was featured at a high-level event hosted by the Italian Ministry of Interior in Rome, bringing together representatives from EUROPOL, CEA-List, law enforcement agencies, and international participants.
The session focused on demonstrating the operational impact of a selection of STARLIGHT-developed AI tools, including technologies for crowd behaviour analysis, object and logo recognition, synthetic face generation, geo-spatial crime prediction, and migration trend forecasting.
From 8th to 10th April 2025, the STARLIGHT consortium reunited in Palermo for its 42-month General Assembly, hosted by project partner Engineering Group. This three-day meeting brought together law enforcement agencies, technical experts, researchers, and policymakers to assess project progress, validate AI tools in operational settings, and plan the next steps towards completion.
From 4–6 February 2025, the STARLIGHT team participated in the Europol Industry and Research Days at Europol HQ in The Hague, Netherlands. There we engaged with law enforcement agencies, industry leaders, and researchers to explore advancements in AI, biometrics, and cybersecurity.