Description
Violent ideologies flourish in online communities that sanction extremist content. Communication in such communities includes a variety of modalities, such as text, memes, videos, and podcasts, which collectively radicalise their consumers. In this position paper, we argue that radicalisation is a nascent area for which machine learning and NLP are particularly apt. On the one hand, these technologies could mitigate the harms of human review of extremist content and stand to validate theories of radicalisation. On the other, such communities present an avenue for addressing key challenges in machine learning and NLP technologies, such as temporal distribution shifts and multi-modal alignment.
Related topic
Types of radicalisation and emerging threats
Link
Language:
English
Date:
16-07-2026
Proposed by:
IPS
Source:
Talat et al., 2025