ACL Findings2026
Danger Depends on the Mind: A Theory-of-Mind Grounded Dataset and Model for Context-Dependent Dangerous Speech
Yuanchen Shi, Longyin Zhang, Guodong Zhou, Fang Kong
Abstract
Dangerous speech detection is a well-studied task, but existing approaches typically treat utterances in isolation, relying on binary labels that ignore who is speaking and in what mental state. We formulate a context-dependent variant of this task by grounding it in Theory-of-Mind (ToM). In cognitive science, ToM studies how humans attribute latent mental states-such as emotions, intentions, and actions-to others. We argue that such states are key signals for assessing the risk of an utterance. Building on this view, we construct ToM-DS, a 79Kinstance dataset where each utterance is paired with structured speaker profiles, ToM states (emotion, intent, action), and topic hierarchies. During data construction, we first identify context-dependent sentences and generate diverse safe and dangerous scenarios surrounding them. High-quality annotations are obtained with state-of-the-art LLMs and a multi-stage cross-agent validation pipeline, yielding a comprehensive and reliable resource for contextdependent dangerous speech detection and fine-grained risk level classification. We further propose ToMGuard, a lightweight model with a dynamic ToM attention mechanism that adaptively weighs different mental-state cues. ToMGuard outperforms strong proprietary and open-source LLMs with significantly fewer parameters. Experimental results show that ToMGuard sets a new benchmark for contextdependent dangerous speech detection and risk level classification on ToM-DS. Our dataset and model are publicly available at https:// github.com/FakerBoom/ToMDS-ToMGGuard.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Lineage
- Facilitating Fine-grained Detection of Chinese Toxic Language: Hierarchical Taxonomy, Resources, and Benchmarks cites COLD: A Benchmark for Chinese Offensive Language Detection
- ToMBench: Benchmarking Theory of Mind in Large Language Models cites FANToM: A Benchmark for Stress-testing Machine Theory of Mind in Interactions
Builds on8
- Unsupervised Cross-lingual Representation Learning at ScaleAlexis Conneau, Kartikay Khandelwal, Naman Goyal, Vishrav Chaudhary et al.ACL 2020 · 539 citations
- COLD: A Benchmark for Chinese Offensive Language DetectionJiawen Deng, Jingyan Zhou, Hao Sun, Chujie Zheng et al.EMNLP 2022 · 82 citations
- Facilitating Fine-grained Detection of Chinese Toxic Language: Hierarchical Taxonomy, Resources, and BenchmarksJunyu Lu, Bo Xu, Xiaokun Zhang, Changrong Min et al.ACL 2023 · 25 citations
- FANToM: A Benchmark for Stress-testing Machine Theory of Mind in InteractionsHyunwoo Kim, Melanie Sclar, Xuhui Zhou, Ronan Le Bras et al.EMNLP 2023 · 21 citations
- ToMBench: Benchmarking Theory of Mind in Large Language ModelsZhuang Chen, Jincenzi Wu, Jinfeng Zhou, Bosi Wen et al.ACL 2024 · 6 citations
Related papers
- Views Are My Own, but Also Yours: Benchmarking Theory of Mind Using Common GroundAdil Soubki, John Murzaku, Arash Yousefi Jordehi, Peter Zeng et al.ACL Findings 2024 · 1 citation
- Agentic-ToM: Cognition-Inspired Agentic Processing For Enhancing Theory of Mind ReasoningSneheel Sarangi, Chetan Talele, Hanan SalamEMNLP Findings 2025 · 2 citations
- ToM-Synth: Scaling Robust Theory of Mind in LLMs via 6, 912 Structured Social UnitsGuiyang Hou, Xiang Huang, Shangke Lyu, Yuchuan Wu et al.ACL Findings 2026
- Beyond Context to Cognitive Appraisal: Emotion Reasoning as a Theory of Mind Benchmark for Large Language ModelsGerard Christopher Yeo, Kokil JaidkaACL Findings 2025 · 3 citations
- Machine Theory of Mind Needs Machine ValidationAdil Soubki, Owen RambowACL Findings 2025
