☀️ Good morning. Here's everything that happened in cybersecurity yesterday — in under 5 minutes.
N-able admitted its first N-central fix didn't work and rushed out Hotfix 2 as intruders kept their grip on managed systems; Iran-linked crews widened their assault on U.S. water utilities to New Jersey and Alabama; PortSwigger showed a single line of CSS can break out of an email and hijack the webmail interface across six major providers; two researchers who bought throwaway domains like noreply.net watched hundreds of companies keep mailing them corporate secrets; and a Kansas City researcher demonstrated an on-demand "adversarial pattern" that makes people and cars vanish from surveillance cameras. The thread running through all of it: attackers spent yesterday exploiting the boring, trusted plumbing — not the front door.
🔥 Top Stories
01 — N-able Ships N-central Hotfix 2 as Attackers Keep Their Foothold
Vulnerabilities
N-able has released a second hotfix for its N-central remote monitoring and management (RMM) platform after the first one failed to evict intruders. Two flaws — CVE-2026-18577 and CVE-2026-18556 — were exploited to reach the managed systems sitting behind the platform, and attackers held their position even after the server itself was locked down with Hotfix 1. Microsoft has tied related activity to China-linked Storm-1175, which deployed a previously undocumented ransomware strain, StormEncryptor.
RMM tools are a force multiplier: one compromise fans out to every downstream client, which is exactly why patching the server isn't enough here. Apply Hotfix 2 even if you already installed Hotfix 1, then actively hunt for persistence on the managed endpoints rather than assuming the fix closed the door.
02 — Iran-Linked Water-System Attacks Spread to New Jersey and Alabama
Nation-State / ICS
The campaign against U.S. water and wastewater facilities keeps widening, with New Jersey and Alabama now joining at least a dozen states hit by hackers linked to Iran. The targets aren't exotic: they're Internet-exposed, poorly secured programmable logic controllers (PLCs) that anyone can find and poke at.
That crudeness is the point — these are opportunistic hits on exposed operational technology, not sophisticated ICS intrusions, which is why they scale so easily across small utilities. Get PLCs off the public Internet, change default credentials, and segment OT from IT now. Congress is moving too (see What to Watch).
03 — PortSwigger's CSS Trick Breaks Webmail Defenses Across Six Providers
Vulnerabilities
New research from PortSwigger's Gareth Heyes shows that CSS embedded inside an email can escape its message boundary and manipulate the surrounding webmail interface — demonstrated against Outlook, Gmail, Fastmail, Proton Mail, Yahoo Mail, and AOL Mail. The technique can be used to steal passwords and tokens or to hijack trusted UI actions the user thinks they're taking safely.
This is a rendering-trust problem, not a user-error one: the message you open can quietly rewrite the page around it. Providers are shipping fixes, but until yours confirms it's patched, treat any unexpected login box or "confirm" prompt that appears inside webmail as suspect.
04 — Researchers Bought noreply.net. Companies Keep Emailing Them Secrets
Data Breach / Identity
Two security researchers bought cheap, throwaway-looking domains — including noreply.net and deleteduser.com — and pointed them at email listening services. Hundreds of companies are still firing corporate secrets, password resets, and internal notifications straight into those inboxes, treating the addresses as digital trash cans.
The good news is that this is a configuration problem your team can audit today. Confirm that your "no-reply" senders use domains you actually own and control, and verify that no sensitive workflow routes mail to an address nobody on your side is watching.
05 — 'Adversarial Pattern' Hides People, Faces, and Cars From Cameras
Privacy / AI
A Kansas City security researcher says his algorithm now produces adversarial patterns on demand that hide people, faces, and vehicles from surveillance cameras. He tested the patterns against 11 open-source detection systems and demonstrated one working on a car at Def Con.
It's a vivid reminder that machine-learning detection can be defeated by inputs crafted to exploit the model itself — no exploit code required. Any organization leaning on automated video analytics for physical security should treat the camera's "brain" as an attack surface, not a settled solution.
📊 By The Numbers
2 hotfixes — N-able now requires N-central Hotfix 2 even on systems already patched with Hotfix 1 (CVE-2026-18577, CVE-2026-18556).
6 webmail services — Outlook, Gmail, Fastmail, Proton Mail, Yahoo, and AOL all fell to PortSwigger's CSS boundary-escape technique.
11 detection systems — the number of open-source object detectors the adversarial camouflage pattern defeated in testing.
A dozen-plus U.S. states — now targeted in the Iran-linked water-utility campaign, with New Jersey and Alabama the latest additions.
$300 million a year — proposed federal funding in a new Senate bill to shore up water-sector cybersecurity.
⚡ The Signal
Nothing yesterday looked like a movie hack. There was no dramatic front-door break-in — just attackers quietly abusing the systems everyone already trusts. An RMM console that IT relies on to manage fleets. The CSS engine that renders your email. A "no-reply" domain nobody thought to actually register. A private cellular network a utility uses to reach remote pumps. Even the neural net inside a security camera. Each is infrastructure we've decided not to think about, and each became the way in.
That's the uncomfortable pattern connecting the N-able intrusion, the webmail research, the noreply.net experiment, and the water-utility attacks: the trusted layer is now the target layer. Defenders instinctively harden the perimeter and the crown-jewel apps, but the day's stories all landed a level below that — in the plumbing, the defaults, and the assumptions. When N-able's first fix failed, it wasn't because the patch was wrong; it was because the attackers had already moved into the trusted systems downstream, where nobody was looking.
The practical takeaway is to widen the blast radius you plan for. Patching the N-central server without hunting the managed endpoints leaves the real foothold in place. Securing your mail server while sensitive resets flow to an unowned domain misses the leak entirely. The work this week is less about buying another tool and more about auditing the trusted defaults you've stopped questioning — because that's precisely where yesterday's attackers went.
🔍 What You May Have Missed
Metabase zero-day exploited to breach Framework customer data — A maximum-severity Metabase flaw exploited as a zero-day let attackers pull names, emails, phone numbers, and addresses from Framework and other customers; the blast radius across the analytics platform's users is wide.
OpenAI expands Daybreak and launches a dedicated GPT-5.6-Cyber model — OpenAI rolled out "Red" and "Blue" programs, a cyber-specialized model, and partnerships with 16 vendors — frontier offensive-security capability that now flows only through approved partners, not their clients.
FBI and South Korea warn of Gunra ransomware hitting critical infrastructure — The Gunra ransomware-as-a-service crew is breaching critical-infrastructure organizations through flaws in popular firewall brands; patch and closely monitor your edge devices.
📅 What to Watch
OpenAI pauses work on its Astra model over cyberattack concerns — Internal evaluations found Astra strong enough in agentic coding and offensive security to trigger new "critical"-tier controls, a signal of where frontier-model risk is heading.
Senate Democrats propose $300M a year for water-sector cybersecurity — A direct legislative response to the widening water attacks; watch whether it picks up bipartisan support.
NATO's cyber arm and AI startup AISLE gain authority to issue CVEs — New CVE Numbering Authorities — including an AI-driven flaw finder — could reshape how vulnerabilities get named and tracked.
New passkey attacks recover synced private keys and bypass phishing-resistant MFA — Three separate research efforts defeated passkey protections without breaking the underlying cryptography; watch how vendors respond as passkeys go mainstream.
Stay sharp. Stay ahead.
Till next time,
The CyberSignal Team