A Comparative Evaluation of AI Agent Security Guardrails

This report presents a comparative evaluation of DKnownAI Guard in AI agent security scenarios, benchmarked against three competing products: AWS Bedrock Guardrails, Azure Content Safety, and Lakera Guard. Using human annotation as the ground truth, we assess each guardrail's ability to detect tw...

UNSEEN: A Cross-Stack LLM Unlearning Defense against AR-LLM Social Engineering Attacks

Emerging AR-LLM-based Social Engineering attack e.g., SEAR is at the edge of posing great threats to real-world social life. In such AR-LLM-SE attack, the attacker can leverage AR Augmented Reality glass to capture the image and vocal information of the target, using the LLM to identify the targe...

ai-security-poc

AI Security POC A fully containerised proof-of-concept for te...

Deterministic + Agentic AI: The Architecture Exposure Validation Requires

Few technologies have moved from experimentation to boardroom mandate as quickly as AI. Across industries, leadership teams have embraced its broader potential, and boards, investors, and executives are already pushing organizations to adopt it across operational and security functions. Pentera’s...

Challenges and Future Directions in Agentic Reverse Engineering Systems

Agentic systems built on large language models LLMs are increasingly being used for complex security tasks, including binary reverse engineering RE. Despite recent growth in popularity and capability, these systems continue to face limitations in realistic settings. Cutting-edge systems still fai...

CVE-2026-40217

A flaw was found in LiteLLM. A remote attacker can exploit this flaw by performing bytecode rewriting at the /guardrails/testcustomcode URI. This could lead to arbitrary code execution, allowing the attacker to run malicious code on the affected system. Mitigation Mitigation for this issue is...

CVE-2026-40217

LiteLLM through 2026-04-08 allows remote attackers to execute arbitrary code via bytecode rewriting at the /guardrails/testcustomcode URI...

CVE-2026-40217

LiteLLM through 2026-04-08 allows remote attackers to execute arbitrary code via bytecode rewriting at the /guardrails/testcustomcode URI...

Arbitrary Code Injection

Overview litellm is a Library to easily interface with LLM API providers Affected versions of this package are vulnerable to Arbitrary Code Injection in the /guardrails/testcustomcode endpoint through bytecode rewriting. An attacker can execute arbitrary code by sending specially crafted requests...

CVE-2026-40217

LiteLLM (through 2026-04-08) is vulnerable to remote code execution via bytecode rewriting at /guardrails/test_custom_code. The CVSSv3.1 vector yields a high severity (8.8) with NETWORK attack, LOW privileges, no user interaction required. Affected component is unspecified beyond the URL vector; ...

LiteLLM 安全漏洞

LiteLLM is an open-source application developed by Berri AI. It allows for the invocation of all LLM APIs in the OpenAI format. Versions of LiteLLM dated before April 8, 2026, contain a security vulnerability. This vulnerability stems from the /guardrails/testcustomcode URI, which allows arbitrar...

ShieldNet: Network-Level Guardrails against Emerging Supply-Chain Injections in Agentic Systems

Existing research on LLM agent security mainly focuses on prompt injection and unsafe input/output behaviors. However, as agents increasingly rely on third-party tools and MCP servers, a new class of supply-chain threats has emerged, where malicious behaviors are embedded in seemingly benign tool...

OpenAI Patches ChatGPT Data Exfiltration Flaw and Codex GitHub Token Vulnerability

A previously unknown vulnerability in OpenAI ChatGPT allowed sensitive conversation data to be exfiltrated without user knowledge or consent, according to new findings from Check Point. "A single malicious prompt could turn an otherwise ordinary conversation into a covert exfiltration channel,...

CVE-2026-32128

FastGPT is an AI Agent building platform. In 4.14.7 and earlier, FastGPT's Python Sandbox fastgpt-sandbox includes guardrails intended to prevent file writes static detection + seccomp. These guardrails are bypassable by remapping stdout fd 1 to an arbitrary writable file descriptor using fcntl...

CVE-2026-31992

OpenClaw versions prior to 2026.2.23 contain an allowlist bypass vulnerability in system.run guardrails that allows authenticated operators to execute unintended commands. When /usr/bin/env is allowlisted, attackers can use env -S to bypass policy analysis and execute shell wrapper payloads at...

CVE-2026-32044 OpenClaw < 2026.3.2 - Tar Archive Safety Bypass in Skills Installation

OpenClaw versions prior to 2026.3.2 contain an archive extraction vulnerability in the tar.bz2 installer path that bypasses safety checks enforced on other archive formats. Attackers can craft malicious tar.bz2 skill archives to bypass special-entry blocking and extracted-size guardrails, causing...

GHSA-X742-88JJ-7HV9 Duplicate Advisory: allowlist exec-guard bypass via env -S

Duplicate Advisory This advisory has been withdrawn because it is a duplicate of GHSA-48wf-g7cp-gr3m. This link is maintained to preserve external references. Original Description OpenClaw versions prior to 2026.2.23 contain an allowlist bypass vulnerability in system.run guardrails that allows...

Duplicate Advisory: allowlist exec-guard bypass via env -S

Duplicate Advisory This advisory has been withdrawn because it is a duplicate of GHSA-48wf-g7cp-gr3m. This link is maintained to preserve external references. Original Description OpenClaw versions prior to 2026.2.23 contain an allowlist bypass vulnerability in system.run guardrails that allows...

CVE-2026-31992

OpenClaw versions prior to 2026.2.23 contain an allowlist bypass vulnerability in system.run guardrails that allows authenticated operators to execute unintended commands. When /usr/bin/env is allowlisted, attackers can use env -S to bypass policy analysis and execute shell wrapper payloads at...

CVE-2026-31992 OpenClaw < 2026.2.23 - Allowlist Exec-Guard Bypass via env -S

OpenClaw versions prior to 2026.2.23 contain an allowlist bypass vulnerability in system.run guardrails that allows authenticated operators to execute unintended commands. When /usr/bin/env is allowlisted, attackers can use env -S to bypass policy analysis and execute shell wrapper payloads at...