Weekly Cyber Threats Recap: N-Able, Chrome, Cars, Fridges & AI-Powered Attacks Surge

This week’s cybersecurity headlines paint a stark picture: from four fresh zero-days in N-Able tools and a Windows privilege escalation bug, to exposed car controls and compromised fridges. Across the board, the recurring theme is this—defenses meant to protect us continue to misfire as attackers exploit overlooked flaws, weak defaults, or even fundamental trust mechanisms.

🚨 Key Incidents This Week

N-Able servers hit with 4 zero-days delivering crypto miners. Vulnerabilities dubbed 4EverAble2Shell, part of N-Central servers by N-Able, have been actively exploited to drop cryptocurrency mining malware. Huntback estimates 21 servers across countries including the U.S., UK, Germany, France, Canada, Sweden, and South Africa have been hit. Fixes were expected from N-Able today.

Windows zero-day SignDrop flips defense into privilege exploit. A local privilege-escalation bug called SignDrop, discovered by Windows researcher Abdelhamid Naceri, abuses the Common Log File System (CLFS) feature. It allows attackers to force CLFS to generate signed files (.cnpf), with SYSTEM privileges, writable by the attacker—resulting in full SYSTEM-level access. Microsoft is investigating. Meanwhile, a similar flaw—BigDiskBuster—was also disclosed, focusing on disruption of Microsoft Defender.

Default credentials expose car controls. Researchers found two automakers still rely on a default administrative console password, enabling remote control—unlocking doors, flashing lights, and sounding horns. The accounts use ForgeRock login systems. Both companies patched the issue roughly two hours after being notified.

Chrome patch addresses actively exploited XML use-after-free bug. Google pushed Chrome updates to fix a severe CVE-2026-125350 vulnerability, actively exploited in the wild. The bug, located in the XML component, is now patched in versions 157.0.8103.50/.51 for Windows and macOS, and 157.0.8103.50 for Linux.

Meta slapped with €805 million fine in Italy over privacy lapse. Italy’s data protection regulator penalized Meta for failing to alert Android users that the Facebook app stored and accessed photos and videos without explicit consent. Since 2023 this feature had operated by default. Meta faces a 60-day window to change defaults, add opt-in consent, and is appealing.

🔍 Other Threats & Emerging Trends

Swarm storage clients compromised. The @ethersphere/bee-js JavaScript client and Swarm Desktop were hijacked via a post-install hook in bee-js version 13.3.0 to steal node wallet keys and install backdoors (Go remote shell on Linux/macOS; hidden ScreenConnect on Windows). The malicious code was signed using GitHub Actions, apparently tied to compromised maintainers’ accounts. This attack has been attributed to a campaign called SleepyDuck (or EtherDuck).

Internet-exposed fridges reveal horrifying RCE risks. Researchers found six pre-authentication remote code execution bugs in Danfoss AK-SM800A refrigerator controllers. An attacker can take over devices entirely, even after reset, or render them permanently unrecoverable—targets include exposed units across Europe. Firmware lacks secure-boot protections.

AI accelerates discovery of vulnerabilities. Using AI tools, analysts examined ten popular TCP/IP stacks in just 90 hours—discovering 99 vulnerabilities (66 new), with over half rated critical or high severity. The cost? Less than $1,000. AI helped rediscover many prior bugs and drastically lowered time and budget barriers. Separately, Google’s Project Zero will halve its disclosure time: from 90 days now, to 60 days beginning February 2, 2027, and 30 days from January 3, 2028, driven by faster bug discovery and exploitation.

Other key alerts:

  • New malware Phexia uses macOS-focused lures and smart contracts for command-and-control fallback mechanisms.
  • Agentic ransomware — where AI agents handle hands-on tasks — continues to mature as operators shift tactics.
  • An active campaign is targeting Microsoft SQL Servers via brute-forcing defaults or weak credentials, delivering malware like XMRig and Gh0st RAT.

Driving home the point: across every sector, an overlap of misconfigurations, default credentials, and inherited technical debt is fueling breaches. Even as attack chains grow sophisticated, many still succeed because basic safeguards were ignored or misapplied.

What this means: Organizations need a dual focus—patching aggressively and auditing fundamentals. Most tools and standards assume certain defaults are secure; this week’s stories show what happens when they’re not. As AI continues slashing discovery, vendors will need lightning-fast response times—and consumers should demand transparency around what’s default and what requires extra configuration. Ignoring the basics can no longer be an option.”