This AI Agent Is Designed to Not Go Rogue

The new open source project IronCurtain uses a unique method to secure and constrain AI assistant agents before they flip your digital life upside down...

LLMs Generate Predictable Passwords

LLMs are bad at generating passwords: There are strong noticeable patterns among these 50 passwords that can be seen easily: All of the passwords start with a letter, usually uppercase G, almost always followed by the digit 7. Character choices are highly uneven ­ for example, L , 9, m, 2, $ and...

EUVD-2026-8617

@enclave-vm/core is vulnerable to Sandbox Escape...

PT-2026-21851

Name of the Vulnerable Software and Affected Versions Enclave versions prior to 2.11.1 Description Enclave is a secure JavaScript sandbox used for safe AI agent code execution. A critical security flaw exists in versions prior to 2.11.1, allowing an attacker to escape the sandbox boundaries and...

Security Insights Where Work Happens: Notion Custom Agents + Wiz MCP

Bring Wiz cloud security insights into your Notion workspace with Custom Agents — enabling automated reporting, investigation, and security workflows where teams already work...

CIBER: A Comprehensive Benchmark for Security Evaluation of Code Interpreter Agents

LLM-based code interpreter agents are increasingly deployed in critical workflows, yet their robustness against risks introduced by their code execution capabilities remains underexplored. Existing benchmarks are limited to static datasets or simulated environments, failing to capture the securit...

Understanding Human-AI Collaboration in Cybersecurity Competitions

Capture-the-Flag CTF competitions are increasingly becoming a testbed for evaluating AI capabilities at solving security tasks, due to the controlled environments and objective success criteria. Existing evaluations have focused on how successful AI is at solving CTF challenges in isolation from...

Agents of Chaos

We report an exploratory red-teaming study of autonomous language-model-powered agents deployed in a live laboratory environment with persistent memory, email accounts, Discord access, file systems, and shell execution. Over a two-week period, twenty AI researchers interacted with the agents unde...

Skill-Inject: Measuring Agent Vulnerability to Skill File Attacks

LLM agents are evolving rapidly, powered by code execution, tools, and the recently introduced agent skills feature. Skills allow users to extend LLM applications with specialized third-party code, knowledge, and instructions. Although this can extend agent capabilities to new domains, it creates...

CVE-2026-27004

OpenClaw is a personal AI assistant. Prior to version 2026.2.15, in some shared-agent deployments, OpenClaw session tools sessionslist, sessionshistory, sessionssend allowed broader session targeting than some operators intended. This is primarily a configuration/visibility-scoping issue in...

The vulnerability of AMD’s graphics processor microprogramming software lies in the improper isolation of common resources between trusted and untrusted agents, allowing attackers to compromise the integrity of protected information.

The vulnerability of AMD’s microprogrammed graphics processor software is related to the improper isolation of common resources between trusted and untrusted agents. Exploiting this vulnerability could allow an attacker to compromise the integrity of the protected information...

CVE-2026-27004

CVE-2026-27004 concerns OpenClaw, an open-source personal AI assistant. In versions prior to 2026.2.15, the issue arises in multi-user/shared-agent deployments where session tools (sessions_list, sessions_history, sessions_send) could expose transcript content across peer sessions due to insuffic...

CVE-2026-27004 OpenClaw session tool visibility hardening and Telegram webhook secret fallback

OpenClaw is a personal AI assistant. Prior to version 2026.2.15, in some shared-agent deployments, OpenClaw session tools sessionslist, sessionshistory, sessionssend allowed broader session targeting than some operators intended. This is primarily a configuration/visibility-scoping issue in...

OpenClaw session tool visibility hardening and Telegram webhook secret fallback

Vulnerability In some shared-agent deployments, OpenClaw session tools sessionslist, sessionshistory, sessionssend allowed broader session targeting than some operators intended. This is primarily a configuration/visibility-scoping issue in multi-user environments where peers are not equally...

From Clawdbot to OpenClaw: Practical Lessons in Building Secure Agents

...

Recursive Language Models for Jailbreak Detection: A Procedural Defense for Tool-Augmented Agents

Jailbreak prompts are a practical and evolving threat to large language models LLMs, particularly in agentic systems that execute tools over untrusted content. Many attacks exploit long-context hiding, semantic camouflage, and lightweight obfuscations that can evade single-pass guardrails. We...

RHSA-2026:2758 Red Hat Security Advisory: resource-agents security update

Bulletin has no description...

RHSA-2026:2712 Red Hat Security Advisory: resource-agents security update

Bulletin has no description...

Important: Red Hat Security Advisory: resource-agents security update

An update for resource-agents is now available for Red Hat Enterprise Linux 7 Extended Lifecycle Support. Red Hat Product Security has rated this update as having a security impact of Important. A Common Vulnerability Scoring System CVSS base score, which gives a detailed severity rating, is...

Important: Red Hat Security Advisory: resource-agents security update

An update for resource-agents is now available for Red Hat Enterprise Linux 8.4 Advanced Mission Critical Update Support and Red Hat Enterprise Linux 8.4 Extended Update Support Long-Life Add-On. Red Hat Product Security has rated this update as having a security impact of Important. A Common...