Docker Sandboxes Flaw on macOS Lets Guest Code Access Host Files
A critical Docker Sandboxes vulnerability on macOS could let malicious guest code escape the shared project directory and read or modify host files.
Security & Privacy
Breaches, offensive and defensive AI, and surveillance.
Attackers Still Slip Through MFA Gaps on VPNs and RDP
A new analysis shows that stolen passwords remain a viable entry point wherever multi-factor authentication is not enforced, especially on VPNs, RDP, and Windows systems.
Ivanti warns of actively exploited Connect Secure zero-day
Ivanti is urging customers to patch a zero-day in its Connect Secure remote access gateway that is already being exploited in attacks.
SAP Patches Maximum-Severity OAuth2 SSRF Flaw in Cloud SDK for Python
SAP's October 2026 security update fixes 20 vulnerabilities, including a maximum-severity OAuth2 server-side request forgery issue in the Cloud SDK for Python.
NinjaOne Adds Agentic AI Technician With Governed MCP Access
Katana AI and the NinjaOne MCP server aim to speed IT troubleshooting while keeping AI actions under existing role-based permissions.
Vercel Sandbox Flaw Could Let Rogue AI Agents Exfiltrate Data
New research finds a certificate-handling weakness in Vercel Sandbox that could let malicious AI agents bypass network restrictions and leak sensitive data.
Meta: GitHub Breach Exposed Code and Data of 1,500
The incident highlights the risk of third-party code repositories as Meta confirms exposure of proprietary code and personal data for 1,500 individuals, mostly US employees.
XWorm RAT Adds Surveillance Plugins for Audio and Browser Sessions
The remote access trojan has expanded its plugin set to turn infected machines into live surveillance tools.
AI Speeds Up Familiar Attacks, Shrinking Defender Response Time
IBM's Steve Wallin says AI is turning known cyberattack methods into faster, harder-to-defend threats that leave enterprises with less time to react.
Bitdefender AIVC Maps Shadow AI on Endpoints
A new Bitdefender layer inventories AI services, coding assistants, local runtimes, and MCP connectors that standard software inventory misses.
MarkLogic Flaws Let Delegated Managers Gain Persistent Admin Control
Two critical flaws in Progress Software's MarkLogic Server could let low-privileged delegated managers seize full administrative control, with fixes available in new releases.
ThreatsDay: Attackers Target AI Tools, Exposed Services, and Old Bugs
A weekly security roundup shows attackers finding new keys in AI tools, exposed services, and other weak points while defenders keep creating new storage spots.
RidgeOne offers on-prem autonomous pentesting for mid-sized organizations
The new appliance keeps AI-driven security testing data within customer environments, addressing data residency concerns.
GTA 6 Leaker Ties Full Game Release to $30M Crypto Market Cap
A hacker claims to possess a playable GTA 6 build and will release it if a cryptocurrency token reaches a $30 million market cap.
npm Trusted Publishing flaw exposed 1,500+ packages to takeover
A flaw in npm's Trusted Publishing let attackers publish malicious versions of packages by claiming abandoned GitHub repository names, affecting more than 1,500 packages.
Adobe Patches 140 Flaws Across 10 Products
The update addresses vulnerabilities in widely used tools including Photoshop, ColdFusion, and Commerce.
International raid on West African scam compounds frees 100 trafficking victims
A joint operation by Ghana, Nigeria, UK and US agencies disrupted fraud networks and highlighted tech platforms' dual role in fighting and profiting from scams.
AI agent regrades most high-severity cloud findings, study finds
New research shows that most high-severity cloud security alerts change severity once an AI agent checks the environment, with most moving down.