Leveraging Large Language Models for Trustworthiness Assessment of Web Applications

The widespread adoption of web applications has made their security a critical concern and has increased the need for systematic ways to assess whether they can be considered trustworthy. However, "trust" assessment remains an open problem as existing techniques primarily focus on detecting known...

TreeTeaming: Autonomous Red-Teaming of Vision-Language Models Via Hierarchical Strategy Exploration

The rapid advancement of Vision-Language Models VLMs has brought their safety vulnerabilities into sharp focus. However, existing red teaming methods are fundamentally constrained by an inherent linear exploration paradigm, confining them to optimizing within a predefined strategy set and...

Towards Leveraging LLMs to Generate Abstract Penetration Test Cases from Software Architecture

Software architecture models capture early design decisions that strongly influence system quality attributes, including security. However, architecture-level security assessment and feedback are often absent in practice, allowing security weaknesses to propagate into later phases of the software...

Not All Tokens Are Created Equal: Query-Efficient Jailbreak Fuzzing for LLMs

Large Language ModelsLLMs are widely deployed, yet are vulnerable to jailbreak prompts that elicit policy-violating outputs. Although prior studies have uncovered these risks, they typically treat all tokens as equally important during prompt mutation, overlooking the varying contributions of...

CVE-2026-4555 D-Link DIR-513 boa formEasySetTimezone memory corruption

A weakness has been identified in D-Link DIR-513 1.10. The impacted element is the function formEasySetTimezone of the file /goform/formEasySetTimezone of the component boa. This manipulation of the argument curTime causes stack-based buffer overflow. The attack can be initiated remotely. The...

T-MAP: Red-Teaming LLM Agents with Trajectory-Aware Evolutionary Search

While prior red-teaming efforts have focused on eliciting harmful text outputs from large language models LLMs, such approaches fail to capture agent-specific vulnerabilities that emerge through multi-step tool execution, particularly in rapidly growing ecosystems such as the Model Context Protoc...

CVE-2026-32114

Discourse is an open-source discussion platform. Prior to versions 2026.3.0-latest.1, 2026.2.1, and 2026.1.2, there is an Insecure Direct Object Reference IDOR vulnerability that allows any authenticated user to access metadata about AI personas, features, and LLM models by providing their...

CVE-2026-32114

Discourse (open‑source discussion platform) contains an Insecure Direct Object Reference (IDOR) vulnerability. Prior to versions 2026.3.0-latest.1, 2026.2.1, and 2026.1.2, any authenticated user can access metadata about AI personas, features, and LLM models by supplying their identifiers. This m...

CVE-2026-32114 Discourse's unscoped status lookups leak restricted metadata

Discourse is an open-source discussion platform. Prior to versions 2026.3.0-latest.1, 2026.2.1, and 2026.1.2, there is an Insecure Direct Object Reference IDOR vulnerability that allows any authenticated user to access metadata about AI personas, features, and LLM models by providing their...

Security of Binary-Modulated Optical Key Distribution against Quantum-Enhanced Coherent Eavesdropping

Optical key distribution OKD protects the physical layer of communication links by taking advantage of the inherent noise present in the photodetection process. It allows for efficient generation of a shared random key between two distant users which can subsequently be used for cryptographic...

The vulnerability of the PyNcclPipe class in the library for working with Large Language Models (LLMs) like vLLM allows a hacker to execute arbitrary code.

The vulnerability of the PyNcclPipe class in the library for working with Large Language Models LLMs like vLLM is related to deficiencies in the deserialization mechanism. Exploiting this vulnerability allows a remote attacker to execute arbitrary code...

Elasticsearch 8.19.8, 9.1.8 Security Update (ESA-2026-18)

Deserialization of Untrusted Data in Elasticsearch Leading to Remote Code Execution Dependency on Vulnerable Third-Party Component CWE-1395 exists in PyTorch used by the machine learning model loading component in Elasticsearch that can allow an attacker to achieve remote code execution via Objec...

CVE-2026-27068

Improper Neutralization of Input During Web Page Generation 'Cross-site Scripting' vulnerability in Ryan Howard Website LLMs.txt website-llms-txt allows Reflected XSS.This issue affects Website LLMs.txt: from n/a through = 8.2.6...

Stack overflow parsing XML with deeply nested DTD content models

...

Measuring and Exploiting Confirmation Bias in LLM-Assisted Security Code Review

Security code reviews increasingly rely on systems integrating Large Language Models LLMs, ranging from interactive assistants to autonomous agents in CI/CD pipelines. We study whether confirmation bias i.e., the tendency to favor interpretations that align with prior expectations affects LLM-bas...

PT-2026-26088

A Cross-Site Scripting XSS vulnerability exists in the web-based configuration interface of Zucchetti Axess access control devices, including XA4, X3/X3BIO, X4, X7, and XIO / i-door / i-door+. The vulnerability is caused by improper sanitization of user-supplied input in the dirBrowse parameter o...

PHOENIX CONTACT FL NAT 跨站请求伪造漏洞

PHOENIX CONTACT FL NAT is a series of industrial security gateways developed by PHOENIX CONTACT GmbH in Germany. PHOENIX CONTACT FL NAT has a cross-site request forgeing vulnerability, which originates from the Link Aggregation configuration interface. This vulnerability may allow unverified remo...

CVE-2026-30695

A Cross-Site Scripting (XSS) vulnerability exists in the web-based configuration interface of Zucchetti Axess access control devices, specifically affecting models XA4, X3/X3BIO, X4, X7, and XIO / i-door / i-door+ . The flaw is triggered by the improper sanitization of user-supplied input in the ...

Security Assessment and Mitigation Strategies for Large Language Models: A Comprehensive Defensive Framework

Large Language Models increasingly power critical infrastructure from healthcare to finance, yet their vulnerability to adversarial manipulation threatens system integrity and user safety. Despite growing deployment, no comprehensive comparative security assessment exists across major LLM...

The vulnerability of Flowise Cloud, a software platform for creating user interfaces based on Language Models (LLMs), allows a hacker to gain access to confidential information.

The vulnerability of Flowise Cloud, a software platform for creating user interfaces based on Language Models LLMs, is related to deficiencies in access control. Exploiting this vulnerability could allow an attacker to gain access to confidential information...