From Controlled to the Wild: Evaluation of Pentesting Agents for the Real-World

AI pentesting agents are increasingly credible as offensive security systems, but current benchmarks still provide limited guidance on which will perform best in real-world targets. Existing evaluation protocols assess and optimize for predefined goals such as capture-the-flag, remote code...

CTFusion: A CTF-Based Benchmark for LLM Agent Evaluation

Recent advances in Large Language Models LLMs have enabled agentic systems for complex, multi-step tasks; cybersecurity is emerging as a prominent application. To evaluate such agents, researchers widely adopt Capture The Flag CTF benchmarks. However, current CTF benchmarks reuse existing...

SkillSafetyBench: Evaluating Agent Safety under Skill-Facing Attack Surfaces

Reusable skills are becoming a common interface for extending large language model agents, packaging procedural guidance with access to files, tools, memory, and execution environments. However, this modularity introduces attack surfaces that are largely missed by existing safety evaluations: eve...

LITMUS: Benchmarking Behavioral Jailbreaks of LLM Agents in Real OS Environments

The rapid proliferation of LLM-based autonomous agents in real operating system environments introduces a new category of safety risk beyond content safety: behavior jailbreak, where an adversary induces an agent to execute dangerous OS-level operations with irreversible consequences. Existing...

PT-2026-41191

Name of the Vulnerable Software and Affected Versions Open WebUI versions prior to 0.9.5 Description An authenticated attacker can perform mass assignment via the 'POST /api/v1/evaluations/feedback' endpoint. This is possible because the FeedbackForm uses a configuration that allows extra fields,...

Oracle Poisoning: Corrupting Knowledge Graphs to Weaponise AI Agent Reasoning

We define Oracle Poisoning, an attack class in which an adversary corrupts a structured knowledge graph that AI agents query at runtime via tool-use protocols, causing incorrect conclusions through correct reasoning. Unlike prompt injection, Oracle Poisoning manipulates the data agents reason ove...

MonitoringBench: Semi-Automated Red-Teaming for Agent Monitoring

We introduce a red-teaming methodology that exposes harder-to-catch attacks for coding-agent monitors, suggesting that current practices may under-elicit attacks and overstate monitor performance. We identify three challenges with current red-teaming. First, mode collapse in attack generation,...

Remote Code Execution (RCE)

dedoc/scramble is vulnerable to Remote Code Execution RCE. The vulnerability is due to unsafe evaluation of user-controlled input during documentation generation, which allows an attacker to execute arbitrary PHP code in the application context...

MT-JailBench: A Modular Benchmark for Understanding Multi-Turn Jailbreak Attacks

Multi-turn jailbreaks exploit the ability of large language models to accumulate and act on conversational context. Instead of stating a harmful request directly, an attacker can gradually steer the conversation toward an unsafe answer. Recent methods demonstrate this risk, but they are usually...

EUVD-2026-28588

SEPPmail Secure Email Gateway before version 15.0.2.1 allows unauthenticated remote code execution in the new GINA UI because an endpoint passes attacker-controlled input from a parameter to Perl's eval...

CVE-2026-44128

SEPPmail Secure Email Gateway before version 15.0.2.1 allows unauthenticated remote code execution in the new GINA UI because an endpoint passes attacker-controlled input from a parameter to Perl's eval...

BIT-JRE-2025-24855

numbers.c in libxslt before 1.1.43 has a use-after-free because, in nested XPath evaluations, an XPath context node can be modified but never restored. This is related to xsltNumberFormatGetValue, xsltEvalXPathPredicate, xsltEvalXPathStringNs, and xsltComputeSortResultInternal...

On the Security of Research Artifacts

Research artifacts are widely shared to support reproducibility, and artifact evaluation AE has become common at many leading conferences. However, AE mainly checks whether artifacts work as claimed and can be reproduced. It largely overlooks potential security risks. Since these artifacts are...

kimai 安全漏洞

Kimai is a web-based, multi-user time tracking application developed by Kimai’s individual developers. Versions of Kimai from 2.27.0 to 2.54.0 contained security vulnerabilities. These vulnerabilities stemmed from the possibility for any ROLEUSER to create tags with formula strings as names using...

CVE-2026-41645: Improper Control of Generation of Code

Nuclei is a vulnerability scanner built on a simple YAML-based DSL. From version 3.0.0 to before version 3.8.0, a vulnerability in Nuclei's expression evaluation engine makes it possible for a malicious target server to inject and execute supported DSL expressions. This happens when HTTP response...

When the Ruler Is Broken: Parsing-Induced Suppression in LLM-Based Security Log Evaluation

LLM-based SOC log classifiers are commonly evaluated using regular-expression pipelines that extract structured fields from free-form model output. We demonstrate that this practice introduces a class of silent, systematic evaluation errors, which we term parsing-induced suppression that can caus...

AI-Driven Security Alert Screening and Alert Fatigue Mitigation in Security Operations Centers: A Comprehensive Survey

Security alert screening is the downstream task of filtering, prioritizing, correlating, and contextualizing alerts for analyst attention in Security Operations Centers. This survey reviews artificial-intelligence-driven alert screening and alert-fatigue mitigation from 2015 to 2026. We synthesiz...

Ubuntu 22.04 LTS / 24.04 LTS / 25.10 / 26.04 LTS : Dynaconf vulnerability (USN-8231-1)

The remote Ubuntu 22.04 LTS / 24.04 LTS / 25.10 / 26.04 LTS host has a package installed that is affected by a vulnerability as referenced in the USN-8231-1 advisory. It was discovered that Dynaconf was incorrectly handling template evaluation in its string resolvers. A remote attacker could...

LCC-LLM: Leveraging Code-Centric Large Language Models for Malware Attribution

LLMs are increasingly explored for malware analysis; however, current LLM-based malware attribution remains limited by unsupported indicators and insufficient code-level grounding for identifying malicious and vulnerable code segments. To address these limitations, this research introduces LCC-LL...

PT-2026-51864

Name of the Vulnerable Software and Affected Versions Linux kernel affected versions not specified Description An issue exists in the netfilter module where the nft ct expect obj eval function allocates an expectation and may call nf ct expect related, but fails to drop its local reference. This...