Insights & Analysis

AI Red Team Missed Probabilistic Jailbreak: What Went Wrong

84 npm Packages Poisoned Despite SLSA L3

On May 11, 2026, attackers published 84 malicious npm packages across 42 @tanstack repositories. Each package carried cryptographically valid SLSA Build Level 3 attestations. If your team relies on SL

Microsoft Removed 73 GitHub Repos After Credential Theft

What Happened Microsoft removed 73 GitHub repositories after malware compromised developer credentials and exposed a PyPI publishing token. The attack targeted the durabletask PyPI token, allowing an

7,000 Langflow Instances Exposed: CVE-2026-5027 Teardown

A path traversal vulnerability in Langflow—an open-source AI development platform with more than 149,000 GitHub stars—is being actively exploited. CVE-2026-5027 allows attackers to write arbitrary fil

Proto6: When Schema Parsing Became an RCE Vector

Cyera s security researchers disclosed six vulnerabilities in protobuf.js that turned schema validation—a process most teams assume is safe—into a remote code execution (RCE) pathway. The vulnerabilit

7,000 Langflow Instances Exposed to RCE

What Happened An unauthenticated remote code execution vulnerability (CVE-2026-5027) in Langflow, an open-source platform for building AI applications, remains unpatched despite active exploitation. T

npm Install Scripts Ran Without Approval for 15 Years

Summary of Events For over 15 years, npm allowed packages to execute arbitrary code during installation through preinstall, install, and postinstall scripts. These scripts ran automatically whenever a

Magento RCE via Extension: CVE-2026-45247 Teardown

Overview of the Vulnerability A critical remote code execution vulnerability in the Mirasvit Full Page Cache Warmer extension for Magento 2 and Adobe Commerce allowed attackers to execute arbitrary co

Hades Malware Hijacked AI Security Tools

What Happened StepSecurity researchers uncovered a malware campaign targeting Python development environments and AI-powered security tools. The malware, named Hades, compromised the C++ library ensma

LiteLLM CVE-2026-42271: Command Injection Chains to Full RCE

What Happened BerriAI s LiteLLM, a proxy layer for managing multiple LLM providers, contained a command injection vulnerability ( CVE-2026-42271 ) that allowed authenticated users to execute arbitrary

CISA Orders Federal Agencies to Patch LiteLLM by June 22

A command injection vulnerability in BerryAI s LiteLLM has been added to CISA s Known Exploited Vulnerabilities catalog and is under active exploitation. CVE-2026-42271 allows attackers to execute arb

ServiceNow's Authentication Bypass: When a Single Config Line Exposed Customer Data

What Happened On June 5, 2026, ServiceNow applied an emergency security update after discovering that attackers had exploited an unauthenticated REST endpoint to query customer instance data. The vuln

Chrome Zero-Day CVE-2026-11645: What Broke and How to Fix It

What Happened On April 27, 2026, security researcher 303f06e3 reported a type confusion vulnerability in Chrome s V8 JavaScript engine to Google. The company confirmed active exploitation and assigned

206 CVEs in One Month: What Broke

Microsoft s January 2025 Patch Tuesday shipped 206 CVEs—a single-month record. Before you blame Microsoft s code quality, consider this: AI-driven vulnerability scanners are finding bugs faster than h

19 PyPI Packages Auto-Executed Credential Theft

On a Tuesday morning, a developer ran pip install on what appeared to be a legitimate Python package. Instantly, before any of their code executed, a credential harvester was active—extracting AWS key