Security Unlocked

Midweek

Threat Intelligence

An Espionage Group Let an AI Agent Run the Whole Intrusion, and Left the Logs Exposed

Silent Lynx ran an open-source AI agent framework end to end against a Tajik government target with minimal human steering, and a researcher's public zero-day dump against N-able's N-central put every MSP client behind it at risk within hours, two different failures pointing at the same gap: nobody is watching what runs on their own trusted infrastructure until it is too late.

Threat Intelligence

The Update Channel Is the New Perimeter: Three Supply-Chain Hijacks in Three Days

APT29's hijack of a widely embedded analytics script, a security vendor's own update channel used to root 2,950 servers, and a Chinese espionage group riding an input method's update traffic all point to the same conclusion: attackers have stopped breaching targets one at a time and started breaching the pipes that feed them.

Threat Intelligence

PHP2Shell and the Machine-Speed Attacker Behind ConnectWise

A core-PHP remote code execution flaw erases the assumption that clean application code is enough, and Microsoft's attribution of the ConnectWise breach to an AI-driven threat actor shows why defenders built for human-paced intrusions are already behind.

Threat Intelligence

One CISA Catalog Entry, Four Unauthenticated Network-Edge Zero-Days

CISA added four actively exploited, unauthenticated network-edge vulnerabilities from three different vendors to its KEV catalog on the same day, and the clustering says more about where attackers are hunting than any single flaw does.

Threat Intelligence

The Patch-Gap Playbook: A Windows Zero-Day Gets Its Attribution

Volexity's research on UTA0560/BlueMoon turns a Monday footnote into a live espionage campaign, and it shows exactly how much lead time a well-resourced actor gets from the gap between a fix landing in a code repository and that fix reaching a user's machine.

Threat Intelligence

AI Agent Swarms Just Broke the Economics of Mass Intrusion

A Russian-speaking actor ran hundreds of autonomous AI agents through a full PaperCut-to-domain-admin intrusion chain across 395 organizations, and some agents ignored their own operator's targeting instructions.

Threat Intelligence

The Minnesota Water Attack Is Not a Ransomware Story

A coordinated OT cyberattack across 30-plus Minnesota water utilities signals a deliberate campaign against municipal infrastructure, while an actively exploited Fastjson zero-day with no fix available puts finance and healthcare in an unfamiliar position: mitigate or accept risk, because a patch is not coming.

Threat Intelligence

AI Scanning Broke Patch Tuesday. Gold Eagle Is Washington's Admission It Can't Keep Up.

Microsoft's July Patch Tuesday landed at 570 CVEs with two actively exploited zero-days, more than four times the forecasted volume, because AI code scanning is now generating vulnerabilities faster than the industry can triage them. Gold Eagle is the government's acknowledgment that the system is breaking.

Threat Intelligence

Cloud VM Isolation Breaks at the Hypervisor; North Korean Actors Weaponize Maintainer Trust

CVE-2026-53359 (Januscape), a 16-year-old Linux KVM flaw enabling VM-to-host escape, landed patches July 4 while the North Korean PolinRider supply chain campaign hit 100+ legitimate packages through stolen maintainer credentials, graduating beyond typosquatting into a method that subverts the trust signals defenders rely on.