_version_ 1866912729502056448
author Lim, Chae-Gyun
Han, Seung-Ho
Byun, EunYoung
Han, Jeongyun
Cho, Soohyun
Joo, Eojin
Kim, Heehyeon
Kim, Sieun
Lee, Juhoon
Lee, Hyunsoo
Lee, Dongkun
Hyeon, Jonghwan
Hwang, Yechan
Lee, Young-Jun
Lee, Kyeongryul
An, Minhyeong
Ahn, Hyunjun
Son, Jeongwoo
Park, Junho
Yoon, Donggyu
Kim, Taehyung
Kim, Jeemin
Choi, Dasom
Lee, Kwangyoung
Lim, Hyunseung
Jung, Yeohyun
Hong, Jongok
Nam, Sooyohn
Park, Joonyoung
Na, Sungmin
Choi, Yubin
Choi, Jeanne
Hong, Yoojin
Jang, Sueun
Seo, Youngseok
Park, Somin
Jo, Seoungung
Chae, Wonhye
Jo, Yeeun
Kim, Eunyoung
Whang, Joyce Jiyoung
Hong, HwaJung
Seering, Joseph
Lee, Uichin
Kim, Juho
Choi, Sunna
Ko, Seokyeon
Kim, Taeho
Kim, Kyunghoon
Ha, Myungsik
Lee, So Jung
Hwang, Jemin
Kwak, JoonHo
Choi, Ho-Jin
author_facet Lim, Chae-Gyun
Han, Seung-Ho
Byun, EunYoung
Han, Jeongyun
Cho, Soohyun
Joo, Eojin
Kim, Heehyeon
Kim, Sieun
Lee, Juhoon
Lee, Hyunsoo
Lee, Dongkun
Hyeon, Jonghwan
Hwang, Yechan
Lee, Young-Jun
Lee, Kyeongryul
An, Minhyeong
Ahn, Hyunjun
Son, Jeongwoo
Park, Junho
Yoon, Donggyu
Kim, Taehyung
Kim, Jeemin
Choi, Dasom
Lee, Kwangyoung
Lim, Hyunseung
Jung, Yeohyun
Hong, Jongok
Nam, Sooyohn
Park, Joonyoung
Na, Sungmin
Choi, Yubin
Choi, Jeanne
Hong, Yoojin
Jang, Sueun
Seo, Youngseok
Park, Somin
Jo, Seoungung
Chae, Wonhye
Jo, Yeeun
Kim, Eunyoung
Whang, Joyce Jiyoung
Hong, HwaJung
Seering, Joseph
Lee, Uichin
Kim, Juho
Choi, Sunna
Ko, Seokyeon
Kim, Taeho
Kim, Kyunghoon
Ha, Myungsik
Lee, So Jung
Hwang, Jemin
Kwak, JoonHo
Choi, Ho-Jin
contents The rapid evolution of generative AI necessitates robust safety evaluations. However, current safety datasets are predominantly English-centric, failing to capture specific risks in non-English, socio-cultural contexts such as Korean, and are often limited to the text modality. To address this gap, we introduce AssurAI, a new quality-controlled Korean multimodal dataset for evaluating the safety of generative AI. First, we define a taxonomy of 35 distinct AI risk factors, adapted from established frameworks by a multidisciplinary expert group to cover both universal harms and relevance to the Korean socio-cultural context. Second, leveraging this taxonomy, we construct and release AssurAI, a large-scale Korean multimodal dataset comprising 11,480 instances across text, image, video, and audio. Third, we apply the rigorous quality control process used to ensure data integrity, featuring a two-phase construction (i.e., expert-led seeding and crowdsourced scaling), triple independent annotation, and an iterative expert red-teaming loop. Our pilot study validates AssurAI's effectiveness in assessing the safety of recent LLMs. We release AssurAI to the public to facilitate the development of safer and more reliable generative AI systems for the Korean community.
format Preprint
id arxiv_https___arxiv_org_abs_2511_20686
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle AssurAI: Experience with Constructing Korean Socio-cultural Datasets to Discover Potential Risks of Generative AI
Lim, Chae-Gyun
Han, Seung-Ho
Byun, EunYoung
Han, Jeongyun
Cho, Soohyun
Joo, Eojin
Kim, Heehyeon
Kim, Sieun
Lee, Juhoon
Lee, Hyunsoo
Lee, Dongkun
Hyeon, Jonghwan
Hwang, Yechan
Lee, Young-Jun
Lee, Kyeongryul
An, Minhyeong
Ahn, Hyunjun
Son, Jeongwoo
Park, Junho
Yoon, Donggyu
Kim, Taehyung
Kim, Jeemin
Choi, Dasom
Lee, Kwangyoung
Lim, Hyunseung
Jung, Yeohyun
Hong, Jongok
Nam, Sooyohn
Park, Joonyoung
Na, Sungmin
Choi, Yubin
Choi, Jeanne
Hong, Yoojin
Jang, Sueun
Seo, Youngseok
Park, Somin
Jo, Seoungung
Chae, Wonhye
Jo, Yeeun
Kim, Eunyoung
Whang, Joyce Jiyoung
Hong, HwaJung
Seering, Joseph
Lee, Uichin
Kim, Juho
Choi, Sunna
Ko, Seokyeon
Kim, Taeho
Kim, Kyunghoon
Ha, Myungsik
Lee, So Jung
Hwang, Jemin
Kwak, JoonHo
Choi, Ho-Jin
Artificial Intelligence
Computers and Society
Machine Learning
The rapid evolution of generative AI necessitates robust safety evaluations. However, current safety datasets are predominantly English-centric, failing to capture specific risks in non-English, socio-cultural contexts such as Korean, and are often limited to the text modality. To address this gap, we introduce AssurAI, a new quality-controlled Korean multimodal dataset for evaluating the safety of generative AI. First, we define a taxonomy of 35 distinct AI risk factors, adapted from established frameworks by a multidisciplinary expert group to cover both universal harms and relevance to the Korean socio-cultural context. Second, leveraging this taxonomy, we construct and release AssurAI, a large-scale Korean multimodal dataset comprising 11,480 instances across text, image, video, and audio. Third, we apply the rigorous quality control process used to ensure data integrity, featuring a two-phase construction (i.e., expert-led seeding and crowdsourced scaling), triple independent annotation, and an iterative expert red-teaming loop. Our pilot study validates AssurAI's effectiveness in assessing the safety of recent LLMs. We release AssurAI to the public to facilitate the development of safer and more reliable generative AI systems for the Korean community.
title AssurAI: Experience with Constructing Korean Socio-cultural Datasets to Discover Potential Risks of Generative AI
topic Artificial Intelligence
Computers and Society
Machine Learning
url https://arxiv.org/abs/2511.20686