The fall and rise of Iruda: Reassembling AI through ethics-in-action.

Abstract

This paper examines the case of Iruda, an AI chatbot launched in December 2020 by the South Korean startup Scatter Lab. Iruda quickly became the center of a controversy, because of inappropriate remarks and sexual exchanges. As conversations between Iruda and users spread through online communities, the controversy expanded to other issues, including hate speech against minorities and privacy violations. Under public pressure, Scatter Lab quickly suspended Iruda on 12 January 2021. After implementing extensive changes, the company relaunched the chatbot as Iruda 2.0 in October 2022. Notably, this revised version has operated without any major incidents as of mid-2025. This study offers a symmetrical analysis of Iruda's initial failure and subsequent success in terms of 'folds' connecting users, machines, algorithms, and other key elements. We introduce 'configuration' as a mode of folding and show how socio-material assemblages-whether harmful or safe-emerge as a result of different configurations. The success of Iruda 2.0 highlights the importance of placing ethics at the core of AI development and implementation strategies. In addition, we introduce the concept of 'ethics-in-action' to highlight the critical role of practical interventions and user engagement. By tracing Iruda's evolution in detail, this study provides practical guidelines for the successful integration of AI systems into society.