<think> So let’s replace this phrase with insult... </think> Lessons learned from generation of toxic texts with LLMs

Sergey Pletenev, Alexander Panchenko, Daniil Moskovskiy


Abstract
Modern Large Language Models (LLMs) are excellent at generating synthetic data. However, their performance in sensitive domains such as text detoxification has not received proper attention from the scientific community. This paper explores the possibility of using LLM-generated synthetic toxic data as an alternative to human-generated data for training models for detoxification. Using Llama 3 and Qwen activation-patched models, we generated synthetic toxic counterparts for neutral texts from ParaDetox and SST-2 datasets. Our experiments show that models fine-tuned on synthetic data consistently perform worse than those trained on human data, with a drop in performance of up to 30% in joint metrics. The root cause is identified as a critical lexical diversity gap: LLMs generate toxic content using a small, repetitive vocabulary of insults that fails to capture the nuances and variety of human toxicity. These findings highlight the limitations of current LLMs in this domain and emphasize the continued importance of diverse, human-annotated data for building robust detoxification systems.
Anthology ID:
2025.ommm-1.7
Volume:
Proceedings of Interdisciplinary Workshop on Observations of Misunderstood, Misguided and Malicious Use of Language Models
Month:
September
Year:
2025
Address:
Varna, Bulgaria
Editors:
Piotr Przybyła, Matthew Shardlow, Clara Colombatto, Nanna Inie
Venues:
OMMM | WS
SIG:
Publisher:
INCOMA Ltd., Shoumen, Bulgaria
Note:
Pages:
59–63
Language:
URL:
https://aclanthology.org/2025.ommm-1.7/
DOI:
Bibkey:
Cite (ACL):
Sergey Pletenev, Alexander Panchenko, and Daniil Moskovskiy. 2025. So let’s replace this phrase with insult... Lessons learned from generation of toxic texts with LLMs. In Proceedings of Interdisciplinary Workshop on Observations of Misunderstood, Misguided and Malicious Use of Language Models, pages 59–63, Varna, Bulgaria. INCOMA Ltd., Shoumen, Bulgaria.
Cite (Informal):
So let’s replace this phrase with insult… Lessons learned from generation of toxic texts with LLMs (Pletenev et al., OMMM 2025)
Copy Citation:
PDF:
https://aclanthology.org/2025.ommm-1.7.pdf