arXiv subjects

Hongyi Lu

Publications and source records attributed to Hongyi Lu.

4 recordsLinked to original sources

arXiv2026

ClawLess: A Security Model of AI Agents

Autonomous AI agents powered by Large Language Models can reason, plan, and execute complex tasks, but their ability to autonomously retrieve information and run code introduces significant security risks. Existing approaches attempt to regulate agent behavior through training or prompting, which does not offer fundamental security guarantees. We present ClawLess, a security framework that enforces formally verified policies on AI agents under a worst-case threat model where the agent itself may be adversarial. ClawLess formalizes a fine-grained security model over system entities, trust scopes, and permissions to express dynamic policies that adapt to agents' runtime behavior. These policies are translated into concrete security rules and enforced through a user-space kernel augmented with BPF-based syscall interception. This approach bridges the formal security model with practical enforcement, ensuring security regardless of the agent's internal design.

cs.CR↗

arXiv2026

GPU-Fuzz: Finding Memory Errors in Deep Learning Frameworks

GPU memory errors are a critical threat to deep learning (DL) frameworks, leading to crashes or even security issues. We introduce GPU-Fuzz, a fuzzer locating these issues efficiently by modeling operator parameters as formal constraints. GPU-Fuzz utilizes a constraint solver to generate test cases that systematically probe error-prone boundary conditions in GPU kernels. Applied to PyTorch, TensorFlow, and PaddlePaddle, we uncovered 13 unknown bugs, demonstrating the effectiveness of GPU-Fuzz in finding memory errors.

cs.CR↗

arXiv2023

MOAT: Towards Safe BPF Kernel Extension

The Linux kernel extensively uses the Berkeley Packet Filter (BPF) to allow user-written BPF applications to execute in the kernel space. The BPF employs a verifier to check the security of user-supplied BPF code statically. Recent attacks show that BPF programs can evade security checks and gain unauthorized access to kernel memory, indicating that the verification process is not flawless. In this paper, we present MOAT, a system that isolates potentially malicious BPF programs using Intel Memory Protection Keys (MPK). Enforcing BPF program isolation with MPK is not straightforward; MOAT is designed to alleviate technical obstacles, such as limited hardware keys and the need to protect a wide variety of BPF helper functions. We implement MOAT on Linux (ver. 6.1.38), and our evaluation shows that MOAT delivers low-cost isolation of BPF programs under mainstream use cases, such as isolating a BPF packet filter with only 3% throughput loss.

cs.CR↗

arXiv2022

DAWN: Matrix Operation-Optimized Algorithm for Shortest Paths Problem on Unweighted Graphs

The shortest paths problem is a fundamental challenge in graph theory, with a broad range of potential applications. The algorithms based on matrix multiplication exhibits excellent parallelism and scalability, but is constrained by high memory consumption and algorithmic complexity. Traditional shortest paths algorithms are limited by priority queues, such as BFS and Dijkstra algorithm, making the improvement of their parallelism a focal issue. We propose a matrix operation-optimized algorithm, which offers improved parallelism, reduced time complexity, and lower memory consumption. The novel algorithm requires $O(E_{wcc}(i))$ and $O(S_{wcc} \cdot E_{wcc})$ times for single-source and all-pairs shortest paths problems, respectively, where $S_{wcc}$ and $E_{wcc}$ denote the number of nodes and edges included in the largest weakly connected component in graph. To evaluate the effectiveness of the novel algorithm, we tested it using graphs from SuiteSparse Matrix Collection and Gunrock benchmark dataset. Our algorithm outperformed the BFS implementations from Gunrock and GAP (the previous state-of-the-art solution), achieving an average speedup of 3.769$\times$ and 9.410$\times$, respectively.

cs.DC↗