Toward Secure LLM Agents: Threat Surfaces, Attacks, Defenses, and Evaluation

Large language model LLM agents are rapidly moving from conversational interfaces to software components that plan, invoke tools, maintain memory, and act on external environments. This transition changes the nature of security risk. In agentic settings, failures are no longer limited to unsafe...

Assessing Automated Prompt Injection Attacks in Agentic Environments

Indirect prompt injection poses a critical threat to LLM agents that interact with untrusted external data, yet automated attack methods--proven effective for jailbreaking--remain underexplored in realistic agentic settings. We present a comprehensive empirical evaluation of automated prompt...

Samba Print Exposure Auditor

This is an exposure assessment tool designed to evaluate whether a Samba print service exposes printer-related RPC functionality to anonymous users through the spoolss endpoint...

Certification of Network Quantum Sensing

The distribution of quantum sensors on quantum networks is a key enabler of quantum technologies in interferometry, gravimetry, timekeeping, biological monitoring, and beyond. Yet, guaranteeing the security of these distributed sensors over noisy, insecure networks remains a formidable challenge...

FreeBSD Security Advisory - FreeBSD-SA-26:31.arm64

FreeBSD Security Advisory - Some Arm CPUs have errata where the ordering of stores and the TLBI+DSB sequence may be incorrect. If one CPU stores to a virtual address while another CPU invalidates the translation for that address, the second CPU's TLBI+DSB may complete before the first CPU's store...

MemVenom: Triggered Poisoning of Multimodal Memories in Web Agents

External memory has become a core component of modern web agents, enabling long-horizon reasoning through the retrieval of past experiences. However, this paradigm introduces a critical vulnerability: malicious content injected into memory can be persistently recalled and repeatedly influence age...

FreeBSD Security Advisory - FreeBSD-SA-26:36.ldns

FreeBSD Security Advisory - When used as a stub resolver over UDP, ldns failed to verify that a received response belonged to the outstanding query. It did not check that the response source address and port matched the query destination, that the transaction ID matched, or that the question...

Context-Based Adversarial Attacks on AI Code Generators: Vulnerability Analysis and Implications

AI-powered code generation systems have transformed software development but introduce critical inference-time security vulnerabilities. This research presents a systematic investigation of context-based adversarial attacks, where strategically crafted contextual inputs, including comments,...

Securing Code Understanding: Detecting Natural Backdoor Vulnerability in Code Language Models

Code Language Models CodeLMs have become integral to software engineering, significantly advancing code intelligence tasks. However, their widespread adoption has raised critical security concerns, particularly regarding susceptibility to backdoor attacks. Recent studies have uncovered naturally...

OpenSSL Security Advisory 20260609

OpenSSL is susceptible to multiple security vulnerabilities. A specially crafted PKCS7 or S/MIME signed message could trigger a use-after-free during PKCS7 signature verification. The Cryptographic Message Services CMS processing fails to perform sufficient input validation on the cipher and tag...

OpenSSL Toolkit 3.6.3

OpenSSL is a robust, fully featured Open Source toolkit implementing the Secure Sockets Layer and Transport Layer Security protocols with full-strength cryptography world-wide. This is the 3.6 release...

OpenSSL Toolkit 4.0.1

OpenSSL is a robust, fully featured Open Source toolkit implementing the Secure Sockets Layer and Transport Layer Security protocols with full-strength cryptography world-wide. This is the 4.0 release...

Layer Order Semantics for Automata-Based Cybersecurity

Layered cybersecurity pipelines transform evidence before they decide on it, and the order of those transformations determines which security facts become visible to which layer. This paper gives layer order a finite-state semantics built from a layer-order automaton, deterministic sequential...

Space Fabric: A Satellite-Enhanced Trusted Execution Architecture

The emergence of decentralized satellite networks and orbital computing platforms creates a pressing need for trust architectures that can operate without physical access to the hardware, without reliance on pre-provisioned vendor secrets, and without dependence on a single manufacturer's...

AI Researchers Must Help Lead Arms Control to Mitigate Military AI Risks

The advancement of AI capabilities compels researchers and the public to be more aware of its potential worldwide impact. A pressing near-term concern is the regulation of military AI applications. Armament manufacturers and defense contractors are increasingly investing in AI capabilities and...

FreeBSD Security Advisory - FreeBSD-SA-26:28.capsicum

FreeBSD Security Advisory - sigqueue2 was marked as permitted in capability mode with the introduction of Capsicum in 2011, but the implementation of kernsigqueue did not include a capability mode check restricting signal delivery to the calling process's own PID...

FreeBSD Security Advisory - FreeBSD-SA-26:27.sound

FreeBSD Security Advisory - The sound4 driver contained two memory-safety errors in its mmap2 support. First, dspmmapsingle validated the requested mapping by checking the sum of the user-supplied offset and length against the buffer size. This addition could overflow, so that a large offset and...

Hiding the Trees in the Forest: Building Network Covert Channels with Hash-Based Covert Carrier Filtering

As an effective anti-censorship mechanism, network covert channels can provide data privacy protection and ensure communication security. However, the covertness of existing network covert channels primarily depends on the secrecy of their covert algorithms. With the increasing depth of research ...

FreeBSD Security Advisory - FreeBSD-SA-26:33.unbound

FreeBSD Security Advisory - Multiple vulnerabilities have been reported in Unbound...

When Discovery Outpaces Remediation: Modeling AI-Accelerated Vulnerability Discovery in Interconnected Systems

Advanced AI systems for code analysis, binary analysis, fuzzing orchestration, and penetration-test planningmay significantly increase the rate at which latent vulnerabilities are discovered. While improved discovery can benefit defenders, it can also overload remediation pipelines and accelerate...

Multi-Domain Check Point IKEv2 Mitigation Script for CVE-2026-50751

A Check Point MDS administrative automation script that iterates through all configured management domains and updates Remote Access VPN global properties to enforce IKEv2-only encryption. The script publishes the resulting configuration changes and is intended as a mitigation measure rather than...

Security-First Approach to API Pipeline Development with Zero-Trust Architecture

Modern enterprises face an accelerating onslaught of API-targeted threats amid a rapidly expanding attack surface. Record volumes of software vulnerabilities continue to accelerate dramatically, with 28,818 CVEs disclosed in 2023 a 38% jump from 2022 and 40,009 CVEs in 2024 another 38% increase,...

Secrets Best Not Shared: DNS Privacy Enhancements for the Constrained IoT

Attackers often identify DNS traffic to disrupt or compromise Internet services. While prior work has focused on encrypting queries using DNS over TLS, HTTPS, or QUIC to counter such attacks, we consider IETF protocols designed for resource-constrained IoT devices and empirically analyze the...

Unveiling Privacy Risks in Multi-Modal Large Language Models: Task-Specific Vulnerabilities and Mitigation Challenges

Privacy risks in text-only Large Language Models LLMs are well studied, particularly their tendency to memorize and leak sensitive information. However, Multi-modal Large Language Models MLLMs, which process both text and images, introduce unique privacy challenges that remain underexplored...

Hardware-Aware QAOA for Honeypot Traffic Partitioning on 100+ Qubit IBM Quantum Processors

Denial-of-service DoS and distributed denial-of-service DDoS mitigation requires separating malicious traffic from benign traffic while minimizing disruption to legitimate users. Prior work proposed mapping honeypot traffic partitioning to a weighted MaxCut problem and solving the resulting graph...

The Chronicles of Radio Frequency Fingerprinting

Radio Frequency Fingerprinting RFF has evolved from an early idea for radar emitter identification into a broad research field for wireless device identification and spectrum monitoring for security. Rather than presenting a conventional literature survey, this work provides a critical historical...

Wazuh 4.14.2 Security Scanner

This Python script is a non-exploitative security scanner designed to test basic responsiveness and message handling behavior of a Wazuh cluster communication endpoint...

Revive Adserver 6.0.6 Security Auditor

This script is a defensive security auditing utility designed to identify exposed services, review configuration weaknesses, and collect security posture information from a Revive Adserver deployment without performing direct exploitation...

ProjeQtor 12.4.3 SQL Injection Validator for Login Endpoints

This Python script is a defensive validation tool designed to identify potential SQL injection indicators in login functionality without modifying database contents or attempting exploitation...

Semantic Multi-Agent Intrusion Detection for IoT:Zero-Day and Adversarial Threats with Risk-Aware Reasoning

The rapid proliferation of Internet of Things IoT devices has enabled unprecedented automation and connectivity, but it has also substantially increased the attack surface, exposing networks to sophisticated cyber threats, including zero-day and adversarial intrusions. Traditional Intrusion...

Model Poisoning against Federated Model Adaptation with Chain of Bit-Flips

Federated Learning FL allows a set of clients to collectively train a global model without sharing local training data. Giving the responsibility of the training to decentralized actors may lead to poisoning attacks: clients controlled by malicious third party potentially poison the training...

UniFi OS Server Unauthenticated Remote Code Execution Chain Detection Script

This tool is a safe detector for the unauthenticated remote code execution chain in UniFi OS Server versions 5.0.6 and below, as disclosed in Ubiquiti Security Advisory Bulletin 064...

Windows Notepad Markdown Link Exposure Test

This Metasploit auxiliary module is a non-exploit, safety-focused research tool designed to generate a Markdown file for analyzing how Windows Notepad handles external links. It creates a controlled test document containing a user-defined URL and stores it locally for inspection...

Windows Notepad WebDAV UNC Reference Markdown File Generator

This Metasploit auxiliary module is a file-format generation tool intended for security testing of a CVE-2026-20841 related to Windows Notepad Markdown handling. It produces a Markdown file containing a UNC WebDAV-style path embedded as a clickable link for behavioral analysis...

Customization under Fire: Plugin Poisoning in Text-To-Image Ecosystem

The prosperity of text-to-image T2I models has fostered a vibrant share-and-play ecosystem centered on Low-Rank Adaptation LoRA plugins, which allow users to customize and share model capabilities with ease. This democratization, however, comes with a hidden but severe security risk. Malicious...

Secure WebSocket Upgrade Handler Auditor for HTTP/HTTPS Services

This Python tool implements a concurrent network auditing framework focused on testing HTTP Upgrade handling behavior, especially WebSocket upgrade negotiation. It connects directly to target servers over TCP or TLS, sends crafted upgrade requests, parses raw HTTP responses, and reports whether t...

The Human Vulnerabilities and Exploits (HVE) Framework

The cybersecurity community has invested over two decades in building standardized frameworks, the Common Vulnerabilities and Exposures CVE system, the Common Vulnerability Scoring System CVSS, and the Common Weakness Enumeration CWE to identify, classify, and remediate threats to digital...

Windows 11 Hardening Auditor for WinRE and BitLocker

This Python-based Windows security utility performs defensive auditing and mitigation tasks focused on Windows Recovery Environment WinRE and CVE-2026-45585, BitLocker protection policies, Secure Boot, TPM configuration, and boot-time execution integrity on Windows 11 systems...

RECON: An LLM-Enhanced Backward Constraint Analysis Framework

While traditional techniques, such as symbolic execution, provide a principled foundation for precise constraint reasoning in program analysis, they struggle to scale to modern software systems mainly due to path explosion, the need for function modeling, and the loss of semantic intent at...

Exploiting Logic Asymmetry in Modern Web Application Firewalls

This research whitepaper demonstrates that even the most modern WAFs remain vulnerable to attacks exploiting logic asymmetry in HTTP protocol processing. Real-world testing on a Weaver Ecology OA system achieved a 100% bypass rate 40/40 test cases, confirming the critical severity of this...

Steganography without Modification: Hidden Communication Via LLM Seeds

We demonstrate that widely deployed Large Language Model LLM inference stacks harbor a steganographic channel that requires no modification to model weights, sampling code, or output distributions. The channel exploits a structural property of deterministic decoding: pseudo-random number generato...

State of Agentic AI Security and Governance

An OWASP white paper analyzing the security, governance, and risk management considerations surrounding agentic AI systems, including autonomous decision-making, tool access, prompt injection, data protection, and organizational oversight. This is version 2.01...

AI Assurance in UK Defence: Challenges in Operationalising JSP 936

This report examines practical challenges in operationalising JSP 936 Part 1 for AI assurance in UK Defence. Using a structured interpretive review of the directive's requirements, the analysis identifies eight thematic challenge areas adequacy of evidence and argument, management of human...

RadKey: An LLM-Guided RF Backscatter System for Through-Wall Keystroke Inference

In today's digitally connected world, keyboards remain the primary interface for inputting sensitive information, making them a persistent target for eavesdropping attacks. While prior keystroke inference techniques have exploited side-channel signals such as acoustics and vibrations, they...

Now You (Still) See Me: Detecting Evasive Steganographic Payloads in LLMs

Large language models can be fine-tuned to encode prompt-borne secrets into fluent, seemingly benign outputs. This creates a steganographic exfiltration risk that is difficult to detect with output-level steganalysis. Recent work proposes mechanistic detection using linear probes that recover the...

AutoSUT: The Environment Semantics Gap in Structured CTI for Adversary Emulation

Structured Cyber Threat Intelligence CTI is increasingly used for adversary emulation, detection evaluation, and cyber range design. However, these workflows still require a target System Under Test SUT whose environment is not fully described by public CTI. We measure how much of that environmen...

Data Agents under Attack: Vulnerabilities in LLM-Driven Analytical Systems

Data agents integrate LLM-driven reasoning with relational data access, executable analytical tools, and multi-step workflow orchestration, making them increasingly central to enterprise analytics. This integration introduces new security vulnerabilities across data resources, database execution,...

NCMD: Benign-Anchored Feature Selection for Imbalanced Network Intrusion Detection

Feature selection is critical for network intrusion detection systems NIDS operating under high-dimensional, highly imbalanced traffic, as found in operational and defense networks. Traditional filter methods rank features using global statistics computed symmetrically across classes and thus fai...

Hardening Agent Benchmarks with Adversarial Hacker-Fixer Loops

Agent benchmarks score submissions with outcome verifiers that are typically hand-written and brittle, leaving them open to reward hacking. We audit 1,968 tasks across five terminal-agent benchmarks and find 323 16% hackable by frontier models given only the task description. This corrupts both...

GitInject: Real-World Prompt Injection Attacks in AI-Powered CI/CD Pipelines

AI-powered agents are increasingly embedded in continuous integration and continuous delivery/deployment CI/CD pipelines to autonomously review pull requests PRs, triage issues, and maintain codebases. These agents ingest untrusted content while operating with elevated repository permissions,...