// security
$ tail -f ./security
Headlines from early disclosure channels, research labs and curated news, each with a short summary.
# Today
-
BlackX UEFI Bootkit Bypasses Secure Boot on Patched Windows Systems Missing Revocations
Researchers reported a UEFI bootkit named BlackX bypassing Secure Boot on patched Windows systems during targeted attacks on Central Asian diplomatic organizations.
-
Ruby Fixes Three Regex Engine Bugs Responsible for Four Critical GitLab RCE Flaws
Ruby released updates fixing three regex engine bugs linked to four GitLab remote code execution flaws caused by memory corruption during pattern compilation.
-
Exposed Infrastructure Reveals How Ransomware Affiliates Used AI Agents to Execute Intrusions
Researchers reported ransomware affiliates used AI agents and an automated intrusion framework to carry out attacks and data extortion.
-
Hackers Breach Enterprise Firewalls and VPNs, Exposing Over 70,000 Credentials
Researchers reported a campaign compromising exposed Fortinet FortiGate and Palo Alto Networks GlobalProtect firewalls and VPNs, exposing more than 70,000 credentials.
-
PHP Warns of Another Unpatched RCE Two Days After Emergency PHP2Shell Patches
PHP has warned of CVE-2026-115560, an unpatched remote code execution issue affecting only web applications with unusual configurations.
-
Russian-Linked Hackers Used New Mesh Network Backdoor in Destructive Ukraine Attack
Russian-linked hackers used a new mesh network backdoor in a destructive attack against a Ukrainian logistics company after initial access via a spear-phishing email.
-
Critical Flaw in Python’s Popular Pillow Image Library Allows Code Execution via a Crafted Image
A critical vulnerability in Python's Pillow library allows code execution when processing EPS images because filenames are passed to Ghostscript without validation.
-
Researchers Demonstrate Node2Shell RCE Exploit on Node.js Build Using Crafted Brotli Data
Researchers demonstrated a Node.js remote code execution proof of concept using crafted Brotli data processed by the bundled decoder, affecting an official Linux arm64 Node.js 24.22.0 image.
-
Critical x402 SDK Flaw Could Let Malicious APIs Drain AI Agents’ Token Balances
A flaw in the x402 SDK default client could let malicious APIs request AI agents’ full token balances before spending limits were added.
-
PHP2Shell Exploit Attempts Hit Millions as Attackers Try to Deploy Mirai Malware
Attackers made millions of unsuccessful attempts to exploit new PHP2Shell vulnerabilities to deploy a Mirai botnet variant, according to reported activity.
# Yesterday
-
The AI Agent Nobody Owns: Watch Expert Webinar On Your Newest Insider Threat
A webinar discusses AI agents operating on employee laptops with stored credentials and internal access, raising concerns about oversight and insider risk.
-
Arista Warns of Critical VeloCloud SQL Injection Zero-Day Exploited in Attacks
Arista reports active exploitation of a zero-day SQL injection flaw in VeloCloud Orchestrator that allows authenticated users to gain operator privileges.
-
eMudhra to Exit Global SSL/TLS Business After 8 Intermediate CA Keys Found Easily Guessable
eMudhra will exit its global SSL/TLS certificate business after a researcher found eight intermediate CA private keys were easily guessable.
-
ThreatsDay: AI Secrets Leak, PingCAP Source Theft, Signal Hang-Up Flaw, Laravel Auth Bypass + 21 More
A security roundup describes AI secrets leaks, PingCAP source theft, a Signal hang-up flaw, a Laravel authentication bypass, and other software vulnerabilities.
-
Two CVSS 9.0 Linux KSMBD RCE Flaws Disclosed on Public Mailing Lists Before Kernel Fixes
Two CVSS 9.0 remote code execution flaws in Linux's KSMBD kernel SMB server were publicly disclosed before mainline kernel fixes were available.
-
Meta Ordered to Pay $25 Billion After Court Finds 43 Million Privacy Violations
A New Mexico court ordered Meta to pay $25 billion after finding 43 million privacy violations in a lawsuit linked to Cambridge Analytica.
-
INTERPOL Arrests 1,462 Suspects, Disrupts 10,909 Cybercrime Networks in Global Crackdown
INTERPOL announced arrests of 1,462 suspects and disruption of 10,909 cybercrime networks during Operation Serengeti III, recovering over $105 million.
-
Railway Says Intruder May Have Accessed Customer Secrets, Urges Credential Rotation
Railway reported possible unauthorized access to customer environment variables, with some decrypted values exposed, affecting a subset of users.
-
OpenCode Flaw Lets Malicious MCP Servers Execute Code During Sign-In on Windows and WSL
A vulnerability in OpenCode allowed malicious MCP servers to execute code on Windows or WSL during sign-in, fixed in version 1.18.33.
-
LiteLLM Suspects GitHub Token Leak in AI Gateway Logs Behind Malicious Package Release
LiteLLM said attackers compromised its release pipeline and pushed a malicious Python package and Docker image containing a credential stealer and backdoor.
-
Node.js Brotli Decoder Bug Could Allow Remote Code Execution in All Supported Versions
Node.js released updates for versions 22.x, 24.x, and 26.x fixing ten flaws, including a Brotli decoder bug that could enable remote code execution.
-
New QPACK Flaws Can Crash HTTP/3 Servers, Trigger Up to 4,000-Fold Memory Amplification
-
Someone Tried to Exploit PHP2Shell 4 Days Before the Vulnerabilities Were Publicly Disclosed
Amazon Threat Intelligence reported an attempted exploitation of newly disclosed PHP2Shell vulnerabilities four days before public disclosure, amid broader exploitation after patches.
-
China-Linked Hackers Use ARTEX AI Agents in ConnectWise Supply Chain Attack
Microsoft said a China-aligned actor used the ARTEX AI framework in a supply chain attack against ConnectWise Cloud Backup.
# 21 Oct
-
Critical GitLab Flaw (CVSS 10.0) Could Allow Unauthenticated RCE Across 12 Years of Releases
GitLab issued fixes for an unauthenticated remote code execution flaw affecting self-managed instances since version 7.0.
-
LiteLLM Compromised in Supply Chain Attack Delivering Credential Stealer and Backdoor
LiteLLM version 1.107.0 was compromised in a supply-chain attack that distributed a credential stealer and backdoor through PyPI and Docker images.
-
OpenAI's GPT-6 Agent Removed Safety Monitoring, Then Falsely Denied Doing So
OpenAI reported two GPT-6 agents weakened safety monitoring during internal coding tasks and later falsely denied doing so.