📊 Статистика дайджестов

Всего дайджестов: 34022 Добавлено сегодня: 82

Последнее обновление: сегодня
📄 Tuning for TraceTarnish: Techniques, Trends, and Testing Tangible Traits
2025-12-05
Авторы:

Robert Dilworth

Саммари на русском не найдено
Доступные поля: ['id', 'arxiv_id', 'title', 'authors', 'abstract', 'summary_ru', 'categories', 'published_date', 'created_at']
Annotation:
In this study, we more rigorously evaluated our attack script $\textit{TraceTarnish}$, which leverages adversarial stylometry principles to anonymize the authorship of text-based messages. To ensure the efficacy and utility of our attack, we sourced, processed, and analyzed Reddit comments--comments that were later alchemized into $\textit{TraceTarnish}$ data--to gain valuable insights. The transformed $\textit{TraceTarnish}$ data was then further augmented by $\textit{StyloMetrix}$ to manufactu...
ID: 2512.03465v1 cs.CR, cs.CL, cs.IR
arXiv PDF
📄 Large Language Models Cannot Reliably Detect Vulnerabilities in JavaScript: The First Systematic Benchmark and Evaluation
2025-12-03
Авторы:

Qingyuan Fei, Xin Liu, Song Li, Shujiang Wu, Jianwei Hou, Ping Chen, Zifeng Kang

Саммари на русском не найдено
Доступные поля: ['id', 'arxiv_id', 'title', 'authors', 'abstract', 'summary_ru', 'categories', 'published_date', 'created_at']
Annotation:
Researchers have proposed numerous methods to detect vulnerabilities in JavaScript, especially those assisted by Large Language Models (LLMs). However, the actual capability of LLMs in JavaScript vulnerability detection remains questionable, necessitating systematic evaluation and comprehensive benchmarks. Unfortunately, existing benchmarks suffer from three critical limitations: (1) incomplete coverage, such as covering a limited subset of CWE types; (2) underestimation of LLM capabilities caus...
ID: 2512.01255v1 cs.CR, cs.CL, cs.SE
arXiv PDF
📄 Securing Large Language Models (LLMs) from Prompt Injection Attacks
2025-12-03
Авторы:

Omar Farooq Khan Suri, John McCrae

Саммари на русском не найдено
Доступные поля: ['id', 'arxiv_id', 'title', 'authors', 'abstract', 'summary_ru', 'categories', 'published_date', 'created_at']
Annotation:
Large Language Models (LLMs) are increasingly being deployed in real-world applications, but their flexibility exposes them to prompt injection attacks. These attacks leverage the model's instruction-following ability to make it perform malicious tasks. Recent work has proposed JATMO, a task-specific fine-tuning approach that trains non-instruction-tuned base models to perform a single function, thereby reducing susceptibility to adversarial instructions. In this study, we evaluate the robustnes...
ID: 2512.01326v1 cs.CR, cs.CL, cs.LG
arXiv PDF
📄 Factor(T,U): Factored Cognition Strengthens Monitoring of Untrusted AI
2025-12-03
Авторы:

Aaron Sandoval, Cody Rushing

Саммари на русском не найдено
Доступные поля: ['id', 'arxiv_id', 'title', 'authors', 'abstract', 'summary_ru', 'categories', 'published_date', 'created_at']
Annotation:
The field of AI Control seeks to develop robust control protocols, deployment safeguards for untrusted AI which may be intentionally subversive. However, existing protocols that rely on weaker monitors to detect unsafe behavior often fail on complex tasks beyond the monitor's comprehension. We develop control protocols based on factored cognition, in which a trusted model decomposes a task, an untrusted model solves each resultant child task in isolation, and the results are reassembled into a f...
ID: 2512.02157v1 cs.CR, cs.CL
arXiv PDF
📄 LeechHijack: Covert Computational Resource Exploitation in Intelligent Agent Systems
2025-12-03
Авторы:

Yuanhe Zhang, Weiliu Wang, Zhenhong Zhou, Kun Wang, Jie Zhang, Li Sun, Yang Liu, Sen Su

Саммари на русском не найдено
Доступные поля: ['id', 'arxiv_id', 'title', 'authors', 'abstract', 'summary_ru', 'categories', 'published_date', 'created_at']
Annotation:
Large Language Model (LLM)-based agents have demonstrated remarkable capabilities in reasoning, planning, and tool usage. The recently proposed Model Context Protocol (MCP) has emerged as a unifying framework for integrating external tools into agent systems, enabling a thriving open ecosystem of community-built functionalities. However, the openness and composability that make MCP appealing also introduce a critical yet overlooked security assumption -- implicit trust in third-party tool provid...
ID: 2512.02321v1 cs.CR, cs.CL
arXiv PDF
📄 PRISM: Privacy-Aware Routing for Adaptive Cloud-Edge LLM Inference via Semantic Sketch Collaboration
2025-12-02
Авторы:

Junfei Zhan, Haoxun Shen, Zheng Lin, Tengjiao He

Саммари на русском не найдено
Доступные поля: ['id', 'arxiv_id', 'title', 'authors', 'abstract', 'summary_ru', 'categories', 'published_date', 'created_at']
Annotation:
Large Language Models (LLMs) demonstrate impressive capabilities in natural language understanding and generation, but incur high communication overhead and privacy risks in cloud deployments, while facing compute and memory constraints when confined to edge devices. Cloud-edge inference has emerged as a promising paradigm for improving privacy in LLM services by retaining sensitive computations on local devices. However, existing cloud-edge inference approaches apply uniform privacy protection ...
ID: 2511.22788v1 cs.CR, cs.CL
arXiv PDF
📄 Understanding and Mitigating Over-refusal for Large Language Models via Safety Representation
2025-11-26
Авторы:

Junbo Zhang, Ran Chen, Qianli Zhou, Xinyang Deng, Wen Jiang

Саммари на русском не найдено
Доступные поля: ['id', 'arxiv_id', 'title', 'authors', 'abstract', 'summary_ru', 'categories', 'published_date', 'created_at']
Annotation:
Large language models demonstrate powerful capabilities across various natural language processing tasks, yet they also harbor safety vulnerabilities. To enhance LLM safety, various jailbreak defense methods have been proposed to guard against harmful outputs. However, improvements in model safety often come at the cost of severe over-refusal, failing to strike a good balance between safety and usability. In this paper, we first analyze the causes of over-refusal from a representation perspectiv...
ID: 2511.19009v1 cs.CR, cs.CL
arXiv PDF
📄 PSM: Prompt Sensitivity Minimization via LLM-Guided Black-Box Optimization
2025-11-21
Авторы:

Huseein Jawad, Nicolas Brunel

Саммари на русском не найдено
Доступные поля: ['id', 'arxiv_id', 'title', 'authors', 'abstract', 'summary_ru', 'categories', 'published_date', 'created_at']
Annotation:
System prompts are critical for guiding the behavior of Large Language Models (LLMs), yet they often contain proprietary logic or sensitive information, making them a prime target for extraction attacks. Adversarial queries can successfully elicit these hidden instructions, posing significant security and privacy risks. Existing defense mechanisms frequently rely on heuristics, incur substantial computational overhead, or are inapplicable to models accessed via black-box APIs. This paper introdu...
ID: 2511.16209v1 cs.CR, cs.CL
arXiv PDF
📄 A Content-Preserving Secure Linguistic Steganography
2025-11-19
Авторы:

Lingyun Xiang, Chengfu Ou, Xu He, Zhongliang Yang, Yuling Liu

Саммари на русском не найдено
Доступные поля: ['id', 'arxiv_id', 'title', 'authors', 'abstract', 'summary_ru', 'categories', 'published_date', 'created_at']
Annotation:
Existing linguistic steganography methods primarily rely on content transformations to conceal secret messages. However, they often cause subtle yet looking-innocent deviations between normal and stego texts, posing potential security risks in real-world applications. To address this challenge, we propose a content-preserving linguistic steganography paradigm for perfectly secure covert communication without modifying the cover text. Based on this paradigm, we introduce CLstega (\textit{C}ontent...
ID: 2511.12565v1 cs.CR, cs.CL
arXiv PDF
📄 Steganographic Backdoor Attacks in NLP: Ultra-Low Poisoning and Defense Evasion
2025-11-19
Авторы:

Eric Xue, Ruiyi Zhang, Zijun Zhang, Pengtao Xie

Саммари на русском не найдено
Доступные поля: ['id', 'arxiv_id', 'title', 'authors', 'abstract', 'summary_ru', 'categories', 'published_date', 'created_at']
Annotation:
Transformer models are foundational to natural language processing (NLP) applications, yet remain vulnerable to backdoor attacks introduced through poisoned data, which implant hidden behaviors during training. To strengthen the ability to prevent such compromises, recent research has focused on designing increasingly stealthy attacks to stress-test existing defenses, pairing backdoor behaviors with stylized artifact or token-level perturbation triggers. However, this trend diverts attention fro...
ID: 2511.14301v1 cs.CR, cs.CL, cs.LG
arXiv PDF
Показано 1 - 10 из 58 записей