10 papers
When Claws Remember but Do Not Tell: Stealthy Memory Injection in Persistent Personal Agents
Yechao Zhang, Shiqian Zhao, Jiawen Zhang +5
Persistent personal agents combine long-term memory with access to users' external environments, enabling personalized foreground assistance and proactive background execution. Thi…
Safety in Self-Evolving LLM Agent Systems: Threats, Amplification, and Case Studies
Ruixiao Lin, Xinhao Deng, Qingming Li +12
Self-evolving LLM agent systems, which autonomously update their model parameters, memory, tools, and architectures, introduce a qualitatively new threat landscape in which adversa…
Beyond Similarity: Trustworthy Memory Search for Personal AI Agents
Jiawen Zhang, Kejia Chen, Jiachen Ma +7
Personal AI agents increasingly rely on long-term memory to provide persistent personalization across sessions. However, existing memory pipelines are largely driven by semantic si…
When Are Teacher Tokens Reliable? Position-Weighted On-Policy Self-Distillation for Reasoning
Xiaogeng Liu, Xinyan Wang, Yingzi Ma +2
On-policy self-distillation (OPSD) trains a student on its own rollouts using a privileged teacher, but its standard objective weights all generated tokens equally, implicitly trea…
VideoSEAL: Mitigating Evidence Misalignment in Agentic Long Video Understanding by Decoupling Answer Authority
Chenhao Qiu, Yechao Zhang, Xin Luo +2
Long video question answering requires locating sparse, time-scattered visual evidence within highly redundant content. Although current MLLMs perform well on short videos, long vi…
Mind Your HEARTBEAT! Claw Background Execution Inherently Enables Silent Memory Pollution
Yechao Zhang, Shiqian Zhao, Jie Zhang +5
We identify a critical security vulnerability in mainstream Claw personal AI agents: untrusted content encountered during heartbeat-driven background execution can silently pollute…
Give Them an Inch and They Will Take a Mile:Understanding and Measuring Caller Identity Confusion in MCP-Based AI Systems
Yuhang Huang, Boyang Ma, Biwei Yan +5
The Model Context Protocol (MCP) is an open and standardized interface that enables large language models (LLMs) to interact with external tools and services, and is increasingly a…
ReasoningBomb: A Stealthy Denial-of-Service Attack by Inducing Pathologically Long Reasoning in Large Reasoning Models
Xiaogeng Liu, Xinyan Wang, Yechao Zhang +5
Large reasoning models (LRMs) extend large language models with explicit multi-step reasoning traces, but this capability introduces a new class of prompt-induced inference-time de…
Low Rank Comes with Low Security: Gradient Assembly Poisoning Attacks against Distributed LoRA-based LLM Systems
Yueyan Dong, Minghui Xu, Qin Hu +5
Low-Rank Adaptation (LoRA) has become a popular solution for fine-tuning large language models (LLMs) in federated settings, dramatically reducing update costs by introducing train…
"MCP Does Not Stand for Misuse Cryptography Protocol": Uncovering Cryptographic Misuse in Model Context Protocol at Scale
Biwei Yan, Yue Zhang, Minghui Xu +5
The Model Context Protocol (MCP) is rapidly emerging as the middleware for LLM-based applications, offering a standardized interface for tool integration. However, its built-in sec…