New from Wiz Research: Critical NGINX flaw. Understand the risk and impact.
Vulnerability DatabaseCVE-2026-125350

CVE-2026-125350: 
vulnerability analysis and mitigation

Overview

CVE-2026-125350 is a use-after-free vulnerability in the XML/XSLT processing component (libxslt) of Google Chrome that allows a remote attacker to execute arbitrary code inside the browser's sandbox via a crafted HTML page. It was reported by Arni Hardarson (Neonix Security) on 2026-10-12 and by an anonymous researcher on 2026-11-06, with Google disclosing and patching it on November 9, 2026. All Google Chrome versions prior to 157.0.8103.50 on Windows, Mac, and Linux are affected. The vulnerability carries a CVSS v3.1 base score of 8.8 (High) (Chrome Releases, GitHub Advisory).

Technical details

The vulnerability is classified as CWE-416 (Use After Free) and resides in Chrome's bundled libxslt library, specifically in the XSLT pattern-cache handling code (xsltTestCompMatchDirect in pattern.c). The root cause is re-entrant evaluation of xsl:number instructions over shared pattern-cache state: when a crafted XSLT stylesheet triggers nested xsl:number evaluation (e.g., via mutually dependent variables or xsl:import chains), libxslt can free and subsequently reuse an XPath node-set object, leaving a stale pointer that is later dereferenced. Exploitation requires no privileges but does require user interaction — a victim must visit a malicious HTML page that loads the crafted XML/XSLT content. Two Chromium bug IDs are associated with this issue: 572578603 and 582442907 (Chrome Releases, GitHub Advisory).

Impact

Successful exploitation allows a remote attacker to execute arbitrary code within Chrome's sandboxed renderer process, which can lead to high confidentiality, integrity, and availability impact within the browser's isolated environment. While the sandbox limits direct host OS compromise, code execution inside the renderer can expose sensitive browser data (cookies, saved credentials, session tokens) and may serve as a stepping stone for sandbox escape via chained vulnerabilities. The vulnerability affects Chrome across Windows, macOS, Linux, and Android, making the attack surface extremely broad (Chrome Releases, GitHub Advisory).

Exploitability

Google confirmed active in-the-wild exploitation at the time of disclosure, making this Chrome's eighth actively exploited zero-day of 2026 (Chrome Releases). CISA added CVE-2026-125350 to its Known Exploited Vulnerabilities (KEV) catalog on November 10, 2026 (CISA KEV). Multiple public PoC repositories appeared on GitHub shortly after disclosure; two were assessed as genuine crash-trigger PoCs (demonstrating renderer crash/denial-of-service without full RCE or sandbox escape), while two others were flagged as fake claims with no functional exploit code (Feedly). The EPSS score is approximately 0.96% (62nd percentile), and no specific threat actor attribution has been publicly confirmed. No weaponized exploit kit integration has been reported.

Exploitation steps

  1. Reconnaissance: Identify targets running Google Chrome versions prior to 157.0.8103.50 on any platform (Windows, macOS, Linux, Android). Unpatched enterprise environments or users who have not restarted Chrome are prime targets.
  2. Craft malicious XSLT/XML payload: Construct an XML document with an associated XSLT stylesheet that triggers re-entrant xsl:number evaluation. This involves crafting patterns such as count="nothing | book[@lang='b'][string($inner)]" with lazily evaluated global variables ($outer, $inner) that call xsl:number recursively, or using xsl:import chains with nested for-each/call-template constructs over document() references to secondary XML files.
  3. Host malicious content: Serve the crafted XML source file (with an <?xml-stylesheet?> PI pointing to the malicious XSLT), along with any secondary XML documents, over an attacker-controlled HTTP server accessible to the victim.
  4. Deliver to victim: Lure the victim into visiting the attacker's URL (e.g., via phishing email, malicious advertisement, or compromised website). The victim's Chrome browser automatically processes the XML/XSLT, triggering the use-after-free in libxslt's pattern cache.
  5. Trigger UAF and achieve code execution: The re-entrant xsl:number evaluation causes libxslt to free and reuse an XPath node-set object, corrupting renderer memory. Public PoCs demonstrate renderer crash (DoS); a weaponized exploit would additionally require heap layout control and object reuse to redirect execution flow to attacker-controlled code within the sandboxed renderer process.
  6. Post-exploitation (within sandbox): With code execution in the renderer, an attacker could access in-browser data (cookies, passwords, session tokens), attempt sandbox escape via a chained kernel or browser process vulnerability, or pivot to further attacks (Feedly, Chrome Releases).

Indicators of compromise

  • Network: Unusual HTTP requests fetching XML documents with associated XSLT stylesheets from unknown or suspicious domains; outbound connections from the Chrome renderer process to unexpected external IPs shortly after loading a web page; requests for multiple XML files (e.g., catalog.xml, second.xml, third.xml, sheet.xsl) from a single origin in rapid succession.
  • Process: Chrome renderer process (chrome.exe / chrome subprocess) crashing unexpectedly ("Aw, Snap!" tab crash) when visiting specific URLs; unusual child processes spawned from the Chrome renderer; Chrome DevTools remote debugging port (9222) being accessed externally in lab/enterprise environments.
  • Logs: Browser crash reports referencing libxslt or XSLT-related stack frames (e.g., xsltTestCompMatchDirect, xsltNumberFormat, pattern.c); repeated renderer crash events in Chrome's crash reporting logs (chrome://crashes) correlated with visits to specific domains.
  • File System: Unexpected files written to the user profile or temp directory by the Chrome renderer process; new browser extensions installed without user action following a suspicious page visit.
  • Behavioral: Chrome tabs crashing immediately upon loading XML/XSLT content from external sources; browser history showing visits to domains serving XML files with complex XSLT stylesheets just prior to anomalous activity (Feedly, GitHub Advisory).

Mitigation and workarounds

Google released the fix in Chrome stable channel version 157.0.8103.50 (and 157.0.8103.51 for Windows/Mac), rolled out on November 9, 2026. Users should update Chrome immediately by navigating to chrome://settings/help and restarting the browser after the update downloads. Microsoft Edge (Chromium-based) users should also apply the corresponding update via the Microsoft Security Response Center advisory. Enterprise administrators should enforce Chrome updates via policy and verify deployment using endpoint management tools. No configuration-based workaround is available; updating to the patched version is the only remediation (Chrome Releases, CISA KEV, Microsoft MSRC).

Community reactions

Google's disclosure explicitly confirmed active exploitation in the wild, calling this Chrome's eighth exploited zero-day of 2026 — a milestone that drew significant commentary from the security community (Chrome Releases). The Register characterized the situation as "Chrome zero-day whack-a-mole" given the high frequency of exploited Chrome vulnerabilities in 2026 (The Register). Security researchers on Mastodon and Bluesky widely shared the advisory, urging immediate patching. CISA, CIS, Canada's CCCS, Hong Kong's HKCERT, India's CERT-In, and the UK's NHS Digital all issued independent advisories reinforcing the urgency of the update. Malwarebytes and BleepingComputer published user-facing guidance emphasizing the need to restart Chrome after updating, noting that many users leave the browser open for extended periods without applying pending updates.

Additional resources

Linux Distribution fix status

Fix availability across major Linux distributions and their releases.

Debian

Affected

bookworm

chromium

Affected

sid

chromium

Affected

trixie

chromium

Affected

Alpine

Fixed

edge

libxslt: 1.1.43-r4

Fixed

v3.21

libxslt: 1.1.42-r3

Fixed

v3.22

libxslt: 1.1.43-r4

Fixed

v3.23

libxslt: 1.1.43-r4

Fixed

Source: This report was generated using AI

Free Vulnerability Assessment

Benchmark your Cloud Security Posture

Evaluate your cloud security practices across 9 security domains to benchmark your risk level and identify gaps in your defenses.

Request assessment

Additional Wiz resources

Get a personalized demo

Ready to see Wiz in action?

"Best User Experience I have ever seen, provides full visibility to cloud workloads."
David EstlickCISO
"Wiz provides a single pane of glass to see what is going on in our cloud environments."
Adam FletcherChief Security Officer
"We know that if Wiz identifies something as critical, it actually is."
Greg PoniatowskiHead of Threat and Vulnerability Management