Manus Agentic AI App Found Vulnerable to Prompt Injection
The flaw shows how AI agents that act on external data can be redirected by hidden malicious instructions.
Security & Privacy
Breaches, offensive and defensive AI, and surveillance.
AI Speeds Up Attackers, but SOC Handoffs Slow Defenders
Security teams face a widening gap between AI-compressed attacker loops and the decision latency caused by handoffs.
Cloudflare Fixes Container Flaw That Exposed Leftover Customer Data
A misconfigured thin-provisioning pool let Cloudflare container customers read other customers' residual disk data; the company patched it and found no signs of real-world exploitation.
Russia's Hybrid War on Europe: Cyber, Drones, and Disinformation
European nations supporting Ukraine are facing a coordinated campaign that blends cyber sabotage, drone attacks, and disinformation.
Microsoft Confirms Windows Updates Can Cause Black Screen Issues
Recent Windows preview updates may leave some users staring at a black screen, and Microsoft has acknowledged the problem.
CISA Election Plan Flags Patching Barriers, Voter Database Threats
A CISA election security plan, requested by Homeland Security Secretary Markwayne Mullin, highlights patching difficulties and voter database attacks as key risks.
September infosec releases focus on AI agent identity and control
Vendors shipped new controls for AI agent discovery, access, and recovery, signaling a shift from credential protection to runtime enforcement.
Policy mandates, not tools, limit retail cyberattack spread
A new survey of retail chains finds that security spending and visibility do not stop attacks from spreading between sites, but a single governance rule does.
AI Agent Guardrails Must Sit at Execution Layer, Not in Prompts
Coralogix CEO Ariel Assaraf argues that system prompts can't enforce boundaries, so enterprises need hard controls between agents and their actions.
SentinelOne Wayfinder threat hunting expands to AWS, Azure, and Google Cloud
The expansion gives SentinelOne Wayfinder customers continuous, expert-led threat hunting across cloud control planes, closing gaps attackers exploit to reach data without touching endpoints.
Threat detection dashboards hide coverage gaps, study finds
A new analysis shows that nearly half of detections in an average organization are ineffective, despite appearing healthy on dashboards.
Bad data tops threat hunters' obstacles, SANS survey finds
Half of threat hunters cite data quality or quantity as their biggest barrier, according to the SANS 2026 Threat Hunting Survey.
Microsoft to Deprecate Windows Deployment Services in Next Server Release
The move signals the end of a long-standing network deployment tool, pushing administrators toward modern provisioning methods.
OnePlus Flaws Let Installed Apps Gain Root Without Permissions
Unpatched vulnerabilities in OnePlus and OPPO software let a no-permission app gain full root control of an Android device.
ClickFix Lures on Hacked Ukrainian Sites Drop Psychedelic Stealer
A ClickFix campaign compromised Ukrainian business sites with fake Cloudflare checks to install a new stealer, while researchers also detailed a second ClickFix chain.
AI Coding Tools and Super-Apps Raise New Privacy Threats
Recent disclosures show AI assistants shipping source code to third-party servers and a state-backed super-app silently surveilling users.
Bipartisan bill would make telecom cyber rules voluntary
New legislation would create voluntary cybersecurity best practices and an optional certification for U.S. telecoms after the Salt Typhoon hacks.
x47.c Botnet Uses xAI Grok for Persistence and API Draining
SecurityWeek reports that the x47.c Windows botnet is using xAI's Grok to automate decisions and drain AI APIs.