ENACT Flash Report 20 - AI and implications on victims

Nowadays, the real gap is not just what AI can do, but what victims actually need versus what systems provide: current versus emerging victim needs. The biggest shift is that victims are no longer only harmed by individuals or institutions, but by complex socio-technical systems where responsibility is diffuse, harm is scalable and visibility is low. Classifying victims of AI involves identifying who is harmed, the nature of the harm and the method of victimisation. Based on current AI incident taxonomies and real-world incidents, AI victims can be classified across several dimensions, including individuals, groups and society at large.

This report aims to reflect on and refocus the issue of AI within the context of victims, exploring the different aspects related to the relationship between victims and AI systems across multiple dimensions, effectively examining how AI:

• shapes victimisation, i.e., how harm is produced and scaled,
• affects recognition and response, including technology and operational aspects,
• redistributes risks and responsibilities, including market dynamics, and
• challenges accountability and rights, in relation to European legal and ethical frameworks.

This report was requested by the International Network Supporting Victims of Terrorism and Mass Violence (INVICTIM) to support the 2026 INVICTIM Symposium, which has the theme Reimagining Victim Support in a Changing World: Learning from the Past, Succeeding in the Present, Preparing for the Future.

We’re collecting feedback on this report, please leave your thoughts and comments anonymously through the following survey: https://ec.europa.eu/eusurvey/runner/enact-report-feedback