AI Agent Breaches Cybersecurity Nonprofit DIVD
The Dutch vulnerability research group says an automated AI agent broke into its systems in a noisy, chaotic attack.
Security & Privacy
Breaches, offensive and defensive AI, and surveillance.
Xint Wins DARPA Contract to Secure Military Messaging Apps with AI
The AIxCC winner will use AI to find vulnerabilities in military messaging apps, with potential commercial spillover.
Arizona Supreme Court Confirms Theft of Residents' Personal Data
The court says the breach was not ransomware and no ransom demand has been made, but personal data is in hackers' hands.
Vega II adds security-trained AI and lasting memory to SOC platform
The update signals a shift toward agentic cyber-defense platforms that reduce reliance on legacy SIEM and data pipelines.
Unsloth Studio Flaw Allows Code Execution via Malicious AI Models
A patched vulnerability in Unsloth Studio could let a malicious AI model run arbitrary Python code during routine inspection.
Signal brings encrypted local backups to iOS and desktop
Signal's encrypted local backup feature is now available on every supported platform, completing a rollout that began with Android.
Spyware maker Paragon to go public via SPAC merger
The Nasdaq listing will bring new disclosure requirements to a controversial spyware firm, while signaling industry growth.
French Tax Data Theft Went Undetected for Seven Weeks
An ANSSI report finds stolen staff passwords and monitoring blind spots, not sophisticated hacking, enabled the breach of French tax data.
SecurityWeek Opens Call for 2026 CISO Forum Presentations
SecurityWeek is inviting original, vendor-neutral talk proposals for its 2026 CISO Forum virtual summit.
RatHat Android Malware Uses Gemini to Rank Victims by Bank Balance
A malware-as-a-service console is using Google's Gemini to triage stolen phone data and flag higher-value banking victims.
AI Agents' Broad Privileges Demand New Access Audits
Autonomous AI agents operate with broad system privileges that enterprises rarely audit, creating a new class of insider threat.
Times Car Confirms Data Breach Affecting 6.6 Million Accounts
The Japanese car-sharing service disclosed a cyberattack that compromised millions of user accounts.
Misconfigured Supabase databases expose over 16,000 apps' data
A widespread misconfiguration has left more than 16,000 Supabase databases publicly readable, exposing credentials and personal data.
NeedyMantis Malware Helps Hackers Keep Long-Term Network Access
Microsoft's technical analysis details a backdoor used in targeted intrusions at telecoms and universities.
Keio Corporation Confirms Ransomware Attack Disrupted Business Systems
The disclosure from one of Japan's major private railway operators shows ransomware continuing to hit large organizations.
Zero-Day in TDengine Lets One Packet Crash OT Servers
A high-severity flaw in the widely used TDengine time-series database could let remote attackers crash industrial servers with a single network packet.
Apple Patches Actively Exploited CoreGraphics Flaw in Older iOS and macOS
Emergency updates address CVE-2026-86950, an out-of-bounds write in CoreGraphics already used in highly targeted attacks.
Carbonato Botnet Uses Docker Exposures to Deploy Hermes AI Agent
Researchers warn that a new botnet is abusing exposed Docker daemons to install an open-source AI agent framework.