Skip to main content

Morning Overview

A dangerous unpatched Windows exploit is now public and can hand attackers full control

A security researcher who says he once worked at Microsoft has put working attack code for an unpatched Windows flaw on the open internet, and anyone who can run a program on an affected machine can use it to become the most powerful account on the system. The exploit, called SignDrop, was published in the second week of November, and the November security updates that Microsoft released the same day contained nothing that fixes it.

Its author, Abdelhamid Naceri, who publishes under the name Nightmare Eclipse, claims that roughly 80% of Windows installations are exposed. That estimate rests on a handful of test builds, and no real-world attack has been reported so far.

What SignDrop does with a Windows log file

SignDrop targets the Common Log File System, the Windows component that manages specialized log files. In October 2025 Microsoft added an authentication check to that driver, and the company’s documentation for the feature says its purpose is to ensure the CLFS driver is the author and writer of log files, with cryptographic codes recorded in a reserved section at the end of each file. A log file created or modified by anything else is treated as unsafe to parse.

The exploit turns that safeguard against itself. BleepingComputer reported that the flaw lies in how CLFS creates the companion authentication file, which carries the extension .cnpf. Access checks are correct at the start, but a privileged process that later opens the log can be made to create that file with SYSTEM rights while it inherits the permissions of the attacker’s original file. Naceri described the result in his write-up: “this legit security mitigation can be used against itself and allow an elevation of privileges.”

The Defender trick and the SYSTEM shell

In the published code, the privileged process is the Microsoft Defender antimalware service, according to The Cyber Express. A standard user prepares a CLFS log and container in locations the user controls, then uses object manager symbolic links so that Defender, when it opens the log, creates the authentication file in a location the attacker has chosen. Because that file is created by a SYSTEM process but takes the permissions of the log, the attacker can write to it, and the chain ends in a SYSTEM shell. SYSTEM is the top privilege level in Windows, above any administrator account.

Security Affairs added that the researcher calls an update function in Defender’s antimalware engine to start the sequence, and that Naceri says the same technique could be used to “turn off any antivirus/EDR.” That wider claim is untested. Naceri’s own comment on the underlying design was blunt about where the fault sits: “Everything is done properly by Microsoft here, but the logic allows abuse.”

How far the 80% estimate stretches

Naceri tested SignDrop on Windows 11 26H2, a Windows 11 26H2 Experimental build and Windows Server 2025 Experimental, and estimated that about 80% of Windows installations are vulnerable. The figure is his own guess rather than a measurement. Systems fall within the claimed scope when the CLFS authentication mode, a registry value under HKLM\SYSTEM\CurrentControlSet\Services\CLFS\Authentication, is set to 0 or 1, meaning enforced or learning. Microsoft’s documentation defines value 2 as the mitigation being disabled by an administrator, and the researcher’s checks treat disabled machines as outside the attack surface.

Switching the protection off is not a cure. Both Naceri and Microsoft’s driver documentation warn that without authentication CLFS goes back to parsing tampered logs, which reopens the memory-corruption bugs the check was built to stop.

An independent test softened the picture slightly. The Hacker News reported that Sungyup Nam of 78researchlab said a colleague, Heewon Song of its PurpleHound team, confirmed that the exploit works but not reliably enough for operational use, and that running it changes every Windows consent prompt to show both a standard-user shell and a SYSTEM shell, which makes an attack easy to spot. In the same report, Defender had to be enabled for the published chain to succeed.

Patch status and a researcher with a record

As of the November updates, Microsoft had issued no CVE number, advisory or patch for SignDrop, and the flaw is not listed in the U.S. Cybersecurity and Infrastructure Security Agency’s catalog of known exploited vulnerabilities, The Hacker News said. SecurityWeek reported on Nov. 13 that Microsoft said it had not observed exploitation in the wild at that time.

Naceri has a history of dropping code before vendors can react. Microsoft confirmed his RoguePlanet Defender exploit as CVE-2026-50656 and fixed it the following month, and his ShieldBreak bypass was later assigned CVE-2026-69414, according to The Hacker News. BleepingComputer adds that Microsoft has warned of legal action over “malicious activity causing real harm” to customers after the researcher disclosed zero-days without reporting them privately first.

Until Microsoft publishes a fix, the open question is how quickly an unreliable proof of concept can be turned into something attackers will actually use. The place to watch is Microsoft’s security update guide, where any CVE for SignDrop would first appear.

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.


More from Morning Overview


Morning Overview is reader-supported. Some links in our articles are affiliate links, and we may earn a commission at no extra cost to you. As an Amazon Associate I earn from qualifying purchases. Full disclosure.