Microsoft patched 56 Windows security vulnerabilities, including three new Zero-Day attacks, one of which was actively used.
For more information about the non-security updates released today, you can check out our Windows 11 Cumulative Updates KB5072033 and KB5071417 articles.
The number of bugs in each vulnerability category is detailed below:
- 28 Elevation of Privileges Vulnerability
- 19 remote code execution vulnerabilities
- 4 disclosure vulnerabilities
- 3 Denial of Service Vulnerabilities
- 2 phishing vulnerabilities
A total of 1,275 CVEs were addressed by Microsoft in 2025, according to data collected by Fortra. Tenable’s Satnam Narang said 2025 also marks the second year in a row that the Windows maker has patched more than 1,000 CVEs.
This update adds to the 17 flaws that the tech giant fixed in its Chromium-based Edge browser after releasing the Patch Tuesday update in November 2025. It also includes a phishing vulnerability in Edge for iOS (CVE-2025-62223, CVSS 4.3).
CVE-2025-62221 (CVSS: 7.8), an exploit-after-release vulnerability in the Windows Cloud Files Minifilter driver, which could allow an authorized attacker to locally elevate privileges and gain system permissions is actively exploited. Mini Cloud File Filter is used by OneDrive, Google Drive, iCloud, and others, although being an essential component of Windows, it will still be present on a system with no programs installed.
It is currently unknown how the vulnerability is used in practice and in what context, but for successful exploitation, an attacker must gain access to the vulnerable system in other ways.
According to Mike Walters, president and co-founder of Action1, an attacker could gain access with reduced privileges using methods such as phishing, web browser exploits, or other known remote code execution flaws, and then combine them with CVE-2025-62221 to take control of the host.
With this access, an attacker can deploy kernel components or abuse signed drivers to evade protection and maintain persistence, and can be used as a domain-wide hacking weapon in conjunction with credential theft scenarios.
Exploitation of CVE-2025-62221 resulted in CISA adding it to the Catalog of Known Exploitable Vulnerabilities (KEV).
The remaining two zero-day vulnerabilities are listed below:
- CVE-2025-54100 (CVSS: 7.8): Command injection vulnerability in Windows PowerShell that allows an unauthorized attacker to execute code locally.
“This is a command typo in the way Windows PowerShell handles web content”– stated Alexey Vovk from Action1. “Allows an unauthenticated attacker to execute arbitrary code in the security context of a user executing a crafted PowerShell command such as Invoke-WebRequest”.
The threat becomes significant when this vulnerability is combined with common attack patterns. For example, an attacker could use social engineering to convince a user or administrator to execute a PowerShell snippet via Invoke-WebRequest, allowing the remote server to return crafted content that causes parsing to fail and lead to code execution and implant deployment.
Microsoft has made changes that display a warning when using PowerShell “Invoke-WebRequest” and prompts the user to add -You are in BasicParsing to prevent code execution.
- CVE-2025-64671 (CVSS: 8.4): Command injection vulnerability in GitHub Copilot for JetBrains allows an unauthorized attacker to execute code locally.
Notably, CVE-2025-64671 follows a broader set of security vulnerabilities, collectively called IDEsaster, that were recently disclosed by security researcher Ari Marzouk. The problems arise from adding agent capabilities to the integrated development environment (IDE), exposing new security risks in the process.
These attacks take advantage of immediate injections against the artificial intelligence (AI) agents built into the IDE and combine them with the base layer of the IDE to cause information disclosure or command execution.
“This uses an ‘old’ attack chain that uses a vulnerable tool, so it’s not exactly part of the new IDEsaster attack chain.”Marzuk said. “Specifically, a vulnerable ‘command execution’ tool that allows bypassing a user-configured permission list.”
The vulnerability indicates that it is possible to execute code on affected nodes by causing LLM to execute commands that bypass security barriers and by adding instructions to the user’s “auto-approve” settings.
“This can be achieved through ‘message cross-implementation’, where the message is not modified by the user but by LLM agents that create their own messages based on the contents of files or data received from the Model Context Protocol (MCP) server, which has become popular with agent-based LLMs.”
Source and record: segu-info.com.ar
