3006/2026/WEB 'Multi-modal Large Language Models (LLMs) and AI pipelines'

Following the introduction of large language models (LLMs) and the engineering of prompts via this webinar, CEPOL, together with the Europol Innovation Lab, continues to professionalise analysts and investigators in becoming more efficient and effective in using LLMs. In this webinar we take an additional perspective and focus specifically on using multi-modal LLMs, setting up AI pipelines and when not to use LLMs. This webinar is part of a CEPOL learning and training programme which includes other (recorded) webinars and an onsite course.

By the end of this webinar, participants will be able to:

• List examples of how LLMs can be used for law enforcement investigation and analysis purposes regarding images, videos, audio and translations;
• Give examples of when it may be better not to use LLMs and justify why alternative tools can be more effective;
• Describe how to design an AI pipeline for using multiple AI systems (including LLMs).

EMPACT training activity

Combatting new forms of organised crime requires more cooperation between police and gendarmes, customs officers, border guards, judges, and prosecutors. EU institutions, agencies and Member States work together to tackle organised and serious international crime through a permanent and key instrument: the European Multidisciplinary Platform Against Criminal Threats (EMPACT).

CEPOL plays a crucial role in the EMPACT mechanism through the provision of specialised training and the identification of training needs within each common security threat. Every year, CEPOL provides a comprehensive training package for each common EU crime priority consisting of onsite, online and exchange training activities.

Check more of CEPOL's EMPACT training activities here.