#191 サイバーarxiv (2025-04-28)

本日紹介した論文の一覧 Simplified and Secure MCP Gateways for Enterprise AI Integration http://arxiv.org/abs/2504.19997v1 Securing Agentic AI: A Comprehensive Threat Model and Mitigation Framework for Generative AI Agents http://arxiv.org/abs/2504.19956v1 Securing GenAI Multi-Agent Systems Against Tool Squatting: A Zero Trust Registry-Based Approach http://arxiv.org/abs/2504.19951v1 DeeCLIP: A Robust and Generalizable Transformer-Based Framework for Detecting AI-Generated Images http://arxiv.org/abs/2504.19876v1 The Automation Advantage in AI Red Teaming http://arxiv.org/abs/2504.19855v1 SILENT: A New Lens on Statistics in Software Timing Side Channels http://arxiv.org/abs/2504.19821v1 Prompt Injection Attack to Tool Selection in LLM Agents http://arxiv.org/abs/2504.19793v1 $\texttt{SAGE}$: A Generic Framework for LLM Safety Evaluation http://arxiv.org/abs/2504.19674v1 From Paper Trails to Trust on Tracks: Adding Public Transparency to Railways via zk-SNARKs http://arxiv.org/abs/2504.19640v1 GenPTW: In-Generation Image Watermarking for Provenance Tracing and Tamper Localization http://arxiv.org/abs/2504.19567v1 Metadata-private Messaging without Coordination http://arxiv.org/abs/2504.19566v1 Security Steerability is All You Need http://arxiv.org/abs/2504.19521v1 The Cost of Performance: Breaking ThreadX with Kernel Object Masquerading Attacks http://arxiv.org/abs/2504.19486v1 FCGHunter: Towards Evaluating Robustness of Graph-Based Android Malware Detection http://arxiv.org/abs/2504.19456v1 Provably Secure Public-Key Steganography Based on Admissible Encoding http://arxiv.org/abs/2504.19454v1 JailbreaksOverTime: Detecting Jailbreak Attacks Under Distribution Shift http://arxiv.org/abs/2504.19440v1 GTSD: Generative Text Steganography Based on Diffusion Model http://arxiv.org/abs/2504.19433v1 ChipletQuake: On-die Digital Impedance Sensing for Chiplet and Interposer Verification http://arxiv.org/abs/2504.19418v1 なお、ポッドキャスト内で紹介する内容は、各論文の概要を日本語で解説したもので、論文概要の著作権は論文著者に帰属します。]]

There are no comments yet.

Please log in to write the first comment.