论文研究
arxiv.org ·
超越常规合规：狡猾数据培养大型语言模型的安全警觉
该 arXiv 论文提出“cunning questions”训练方法：这类问题本身未必与安全相关，但包含误导性前提、异常推理或细微矛盾，作者假设学会识别此类推理陷阱可迁移到安全场景。实验显示该训练提升了对分布外越狱攻击的鲁棒性，并增强后续安全微调效果；将其加入现有安全对齐流程后，在九个 backbone-benchmark 组合上把平均 ASR 从 17.40% 降至 15.05%。匹配安全微调后的 trace 分析表明安全判断更可能在有害规划开始前介入，论文还给出关于狡猾数据所学不变性何时可迁移的条件性理论分析。
AI推荐理由 arXiv 新论文（2026-09-16 提交），提出用含误导前提、异常推理的“cunning questions”训练 LLM 的安全警觉性，报告在九个 backbone-benchmark 组合上把平均 ASR 从 17.40% 降到 15.05%，并给出匹配安全微调后的 trace 分析与条件性理论分析。属于有具体方法、数据和局限说明的安全对齐研究，对关注 LLM 安全与对齐的开发者有参考价值；但结论基于作者自评实验，尚未见独立复现，且摘要未披露数据集与基线细节。
72
正文
Computer Science > Artificial Intelligence
arXiv:2609.18515 (cs)
[Submitted on 16 Sep 2026]
Title:Beyond Routine Compliance: Cunning Data Cultivates Safety Vigilance in Large Language Models
Authors:Youjia Wang, Lin Xu, Yang Sun, Yuxiao Lu, Chengfang Fang, Jie Shi
View a PDF of the paper titled Beyond Routine Compliance: Cunning Data Cultivates Safety Vigilance in Large Language Models, by Youjia Wang and 5 other authors
View PDF
HTML (experimental)
Abstract:Safety alignment teaches large language models (LLMs) to recognize harmful requests and reject risky instructions. Yet aligned models can fail when harmful intent is concealed within seemingly benign contexts. Robust safety therefore requires both knowledge of safety boundaries and \textbf{vigilance}: the ability to detect unusual premises, misleading reasoning, and latent risks beneath surface-level semantics. Vigilance requires models to scrutinize a request's underlying intent and assumptions before acting. To cultivate this capability, we introduce \textbf{cunning questions}, which are not necessarily safety-related but contain misleading premises, atypical reasoning, or subtle inconsistencies. We hypothesize that learning to look beyond such reasoning traps can transfer to safety-critical scenarios. Experiments show that Cunning training improves robustness to out-of-distribution jailbreak attacks and strengthens subsequent safety fine-tuning. Furthermore, augmenting an existing state-of-the-art safety alignment pipeline with Cunning establishes a new state of the art across our evaluated settings, reducing mean ASR across nine backbone--benchmark combinations from 17.40\% to 15.05\%. Trace analysis after matched safety fine-tuning suggests that safety judgments are more likely to govern responses before harmful planning begins. A conditional theoretical analysis further characterizes when invariance learned from cunning data can transfer to safety-related inputs. These findings suggest that cunning data can strengthen model vigilance and complement conventional safety alignment.
Comments:
Subjects:
Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
ACM classes:
I.2.7
Cite as:
arXiv:2609.18515 [cs.AI]
(or
arXiv:2609.18515v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2609.18515
Focus to learn more
arXiv-issued DOI via DataCite (pending registration)
Submission history
From: Youjia Wang [view email]
[v1]
Wed, 16 Sep 2026 11:46:48 UTC (1,203 KB)
Full-text links:
Access Paper:
View a PDF of the paper titled Beyond Routine Compliance: Cunning Data Cultivates Safety Vigilance in Large Language Models, by Youjia Wang and 5 other authors
View PDF
HTML (experimental)
TeX Source
view license
Current browse context:
cs.AI
new
|
recent
| 2026-09
Change to browse by:
cs
cs.LG
References & Citations
NASA ADS
Google Scholar
Semantic Scholar
export BibTeX citation
Loading...
BibTeX formatted citation
×
loading...
Data provided by:
Bookmark
Bibliographic Tools
Bibliographic and Citation Tools
Bibliographic Explorer Toggle
Bibliographic Explorer (What is the Explorer?)
Connected Papers Toggle
Connected Papers (What is Connected Papers?)
Litmaps Toggle
Litmaps (What is Litmaps?)
scite.ai Toggle
scite Smart Citations (What are Smart Citations?)
Code, Data, Media
Code, Data and Media Associated with this Article
alphaXiv Toggle
alphaXiv (What is alphaXiv?)
Links to Code Toggle
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub Toggle
DagsHub (What is DagsHub?)
GotitPub Toggle
Gotit.pub (What is GotitPub?)
Huggingface Toggle
Hugging Face (What is Huggingface?)
ScienceCast Toggle
ScienceCast (What is ScienceCast?)
Demos
Demos
Replicate Toggle
Replicate (What is Replicate?)
Spaces Toggle
Hugging Face Spaces (What is Spaces?)
Spaces Toggle
TXYZ.AI (What is TXYZ.AI?)
Related Papers
Recommenders and Search Tools
Link to Influence Flower
Influence Flower (What are Influence Flowers?)
Core recommender toggle
CORE Recommender (What is CORE?)
Author
Venue
Institution
Topic
About arXivLabs
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Which authors of this paper are endorsers? |
Disable MathJax (What is MathJax?)