Description

本日紹介した論文の一覧

• LLM Hypnosis: Exploiting User Feedback for Unauthorized Knowledge
  Injection to All Users
  http://arxiv.org/abs/2507.02850v1

• Visual Contextual Attack: Jailbreaking MLLMs with Image-Driven Context
  Injection
  http://arxiv.org/abs/2507.02844v1

• NVIDIA GPU Confidential Computing Demystified
  http://arxiv.org/abs/2507.02770v1

• Early Signs of Steganographic Capabilities in Frontier LLMs
  http://arxiv.org/abs/2507.02737v1

• Meta SecAlign: A Secure Foundation LLM Against Prompt Injection Attacks
  http://arxiv.org/abs/2507.02735v1

• Quantifying Classifier Utility under Local Differential Privacy
  http://arxiv.org/abs/2507.02727v1

• Control at Stake: Evaluating the Security Landscape of LLM-Driven Email
  Agents
  http://arxiv.org/abs/2507.02699v1

• SAT-BO: Verification Rule Learning and Optimization for FraudTransaction
  Detection
  http://arxiv.org/abs/2507.02635v1

• Access Control Threatened by Quantum Entanglement
  http://arxiv.org/abs/2507.02622v1

• Alleviating Attack Data Scarcity: SCANIA's Experience Towards Enhancing
  In-Vehicle Cyber Security Measures
  http://arxiv.org/abs/2507.02607v1

• De-AntiFake: Rethinking the Protective Perturbations Against Voice
  Cloning Attacks
  http://arxiv.org/abs/2507.02606v1

• Real-Time Monitoring and Transparency in Pizza Production Using IoT and
  Blockchain
  http://arxiv.org/abs/2507.02536v1

• A 10-bit S-box generated by Feistel construction from cellular automata
  http://arxiv.org/abs/2507.02489v1

• Effectively Identifying Wi-Fi Devices through State Transitions
  http://arxiv.org/abs/2507.02478v1

• CyberRAG: An agentic RAG cyber attack classification and reporting tool
  http://arxiv.org/abs/2507.02424v1

• Privacy-preserving Preselection for Face Identification Based on Packing
  http://arxiv.org/abs/2507.02414v1

• Evaluating Language Models For Threat Detection in IoT Security Logs
  http://arxiv.org/abs/2507.02390v1

• PII Jailbreaking in LLMs via Activation Steering Reveals Personal
  Information Leakage
  http://arxiv.org/abs/2507.02332v1

• Rethinking Broken Object Level Authorization Attacks Under Zero Trust
  Principle
  http://arxiv.org/abs/2507.02309v1

• Linearly Homomorphic Ring Signature Scheme over Lattices
  http://arxiv.org/abs/2507.02281v1

• EIM-TRNG: Obfuscating Deep Neural Network Weights with
  Encoding-in-Memory True Random Number Generator via RowHammer
  http://arxiv.org/abs/2507.02206v1

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。