Re-generating hate

This short piece examines how generative AI is becoming embedded in far-right propaganda ecosystems, drawing on examples from the 2023 Dublin riots and the 2024 Southport unrest. Using a media experiment, it traces a re-generative pipeline in which extremist images collected from Telegram and X are first converted into text prompts and then re-created through commercial image models. The results suggest that AI systems do not simply neutralize hateful content: they often sanitize its most explicit markers while preserving, displacing or rearticulating underlying ideological bias. The article argues that this affinity between far-right visual culture and generative AI is not accidental, but rooted in the biases and aesthetic logics built into contemporary commercial models.