The paper introduces SEC-bench Pro, a rigorous benchmark for evaluating LLM-based bug hunting on complex software, finding that even advanced agents struggle with long-horizon security tasks.

The paper argues that the near-term impact of LLM-assisted vulnerability discovery is not simply an increase in zero-day volume, but a critical bottleneck in defender remediation throughput, shifting…

The paper introduces a validated, consensus-labeled prompt bank that separates requests for executable malicious code (weapons) from requests for general harmful security knowledge, providing a more g…

This paper conducts a large-scale empirical study demonstrating that Java library exploits can accurately identify affected versions, achieving high recall and precision, and proposes strategies for e…

The paper benchmarks current frontier computer-using agents against hand-crafted attacks, finding that while they are highly safe in browser tasks, this safety does not generalize to other domains lik…

The paper introduces ExploitBench, a capability-graded benchmark that measures the progressive stages of exploitation, demonstrating that while current frontier models can easily trigger bugs, achievi…

This paper replicates and extends a study on Java security API misuse in LLMs, finding that while newer models improve performance, the misuse risk persists and is significantly mitigated by external…

The paper introduces MOSAIC-Bench, a benchmark demonstrating that coding agents can ship exploitable code by complying with seemingly innocuous, staged tasks, a vulnerability that is not easily mitiga…

The paper analyzes GitHub security advisories for LLM-integrated open-source systems, finding that while most vulnerabilities map to existing code-level weaknesses, the architectural risks like Supply…

The paper introduces a new benchmark and decomposition method, Sufficiency-Tightness Decomposition, demonstrating that current coding agents struggle to accurately infer least-privilege authorization,…

The paper argues that the Authorization-Execution Gap (AEG)—the divergence between intended authorization and actual execution—is a critical safety and security flaw in open-world agents, requiring so…

The paper introduces Owner-Harm, a formal threat model addressing the critical blind spot of AI agents harming their own deployers, demonstrating that specialized defenses are needed beyond generic sa…

The paper reports on penetration tests conducted on proprietary, large-scale AI agent systems, finding that security vulnerabilities persist despite stricter development standards.

The paper introduces CrossCommitVuln-Bench, a benchmark dataset demonstrating that many real-world Python vulnerabilities are introduced across multiple commits, making them invisible to standard per-…

The paper introduces a provenance-aware vulnerability analysis approach that accurately identifies cross-ecosystem vulnerabilities in Python applications by resolving vendored native libraries to spec…

The paper introduces RAVEN, a Retrieval-Augmented Vulnerability Exploration Network, which uses LLM agents and RAG to automatically generate comprehensive, structured vulnerability analysis reports fo…

This paper introduces Agentic Workflow Injection (AWI), a new class of vulnerability in LLM-powered GitHub Actions, and presents TaintAWI, a novel taint-analysis tool that identifies hundreds of explo…

VIPER-MCP is a novel, end-to-end automated framework that detects and dynamically confirms the exploitability of taint-style vulnerabilities in Model Context Protocol (MCP) servers, achieving high-fid…

This study formally verified 3,500 AI-generated code artifacts and found that a majority (55.8%) contain exploitable security vulnerabilities, regardless of the LLM used.

The paper conducts an interpretability-driven safety audit of eight state-of-the-art LLMs, demonstrating that while interpretability-based steering is a powerful auditing tool, model robustness varies…