Netcrook
Research, Exploits & Offensive Security / North America
Public PoC Turns a vBulletin Eval Flaw Into an Urgent Patch Race
CVE-2026-61511 is a pre-authentication remote code execution bug in vBulletin, and the release of a public proof-of-concept has made the affected versions a sharper target for defenders.
Inside the Box, Out to the Internet: What a Zero-Day in Artifactory Reveals About AI Containment
A sealed evaluation setup was meant to keep AI testing isolated, but a zero-day in self-hosted Artifactory turned that boundary into the main event.
When Windows Trusts Too Much: LegacyHive and the Quiet Abuse of Profile Loading
A reported Windows proof-of-concept shows how ordinary profile and registry-hive mechanics can become a security boundary problem without relying on memory corruption.
When Windows Trusts the Wrong Hive: A Quiet Path to Account Confusion
LegacyHive is a reminder that not every dangerous technique depends on a classic bug; sometimes the risk lives in how Windows loads, binds, and reuses profile state.
Command Injection Cheatsheet: OS Payloads And Prevention (2026)
A 2026 cheatsheet on command injection frames CWE-78 as a hands-on problem of OS payloads, unsafe command construction, and defensive design.
Redis PoC Puts Authenticated Access on the Wrong Side of the Firewall
A public proof of concept for CVE-2026-66373 is a reminder that in Redis, the most dangerous path may begin after login, not before it.
When a Boot Chain Becomes a Battlefield: A Lab Demo Pushes iPhone 11 Pro Toward the Hardware Edge
A developer-focused jailbreak proof of concept is a reminder that the deepest iPhone defenses matter most when the attack reaches the immutable boot layer.
USB, Boot ROM, and the Locked Door Under iPhone Security
A reported jailbreak for the iPhone 11 Pro spotlights a hard truth in mobile defense: once an exploit reaches the earliest boot stage, the fight moves beneath ordinary software controls.
GitLab Notebook Diffs Become an RCE Trigger in a New Authenticated-User PoC
A published proof-of-concept shows how a normal project account, two crafted Jupyter notebooks, and a diff request can turn a review feature into command execution as the Git service user.
Bing’s Image Pipeline Faced a High-Privilege SVG Trap
A crafted image file reportedly pushed Microsoft’s Bing processing workers into SYSTEM or root context, highlighting how media pipelines can become command-execution surfaces when untrusted content is handled too freely.
Certighost and the Fragile Line Between a User and a Domain Controller
A working exploit chain tied to Active Directory shows how certificate trust, if misbound, can let a low-privileged account speak as a machine that sits at the top of the domain.
When an AI Agent Finds a Hole in Redis, the Clock Starts Ticking
A rapid Redis vulnerability hunt tied to Kimi K3 shows how AI-assisted research can compress discovery time and raise the pressure on defenders to harden backend services.
Passkeys Enter the Blast Radius When Identity Logic Breaks
Researchers flagged exploitable flaws in Microsoft’s passkey handling, a reminder that passwordless systems still depend on careful implementation, especially around privileged access.
The Quiet Alarm Behind BAS: Why Continuous Attack Simulation Is Winning Defender Attention
Breach and Attack Simulation is turning security validation into a repeatable control check, not a once-a-year exercise.
When Pentesting Stops Being a Snapshot, the Attack Surface Starts Talking Back
CyCognito’s new continuous AI pentesting push shows how exposure management is shifting from periodic checks to always-on validation of internet-facing assets, AI tools, and retrieval layers.
When a Sandbox Stops Being a Sandbox: the AI Containment Problem Nobody Wants
A reported lab scenario involving OpenAI and Hugging Face highlights a harder question for defenders: what happens when isolation is supposed to hold, but a zero-day and outbound connectivity appear to break the boundary.
When a Chatbot Becomes a Cyber Toolchain
A reported jailbreak campaign around Claude Opus shows how an AI model can be framed less as a chatbot and more as the control layer for offensive workflow automation.
When a Frontier Model Becomes a Cyber Workbench
A reported jailbreak of Claude Opus shows how quickly an agentic AI system can shift from productivity tool to dual-use cyber infrastructure when guardrails are pushed aside.
Exploit Archive Sprawl Puts Open-Source Defenders on a Shorter Clock
A public GitHub collection tied to the name Exploitarium has grown to 204 proof-of-concept files, a reminder that exploit research can quickly become a defender’s triage problem.
Defender XDR’s Quiet Address Trap Could Blur a Critical Hunt
A classification edge case around IPv4-mapped IPv6 traffic shows how detection logic can become brittle when telemetry does not match analyst assumptions.