CVE-2026-106329

Executive Summary

CVE-2026-106329 is an incorrect authorization (CWE-863) vulnerability in Google Chrome's FileSystem component affecting versions prior to 155.0.8059.39. A remote attacker can exploit this flaw via a crafted HTML page, leveraging social engineering to trick the victim into interaction, ultimately potentially achieving arbitrary code execution outside the browser sandbox. With a CVSS score of 9.6 (Critical), this vulnerability represents a significant risk to enterprises where users browse untrusted web content, as successful exploitation bypasses Chrome's core security boundary and enables full system compromise.

Technical Analysis

This vulnerability is classified as CWE-863: Incorrect Authorization, located within Chrome's FileSystem implementation. The core issue stems from improper enforcement of authorization checks in the FileSystem component, which governs how the browser interacts with the host filesystem (sandboxed file access, File API operations, Origin Private File System, and related storage backends).

Authorization errors in this context mean that a security decision — typically "is this caller allowed to perform this filesystem operation?" — is made incorrectly. Rather than validating that a request originates from a legitimate, permitted context (correct origin, correct renderer process privilege level, correct sandbox token), the FileSystem component either skips the check, applies it against the wrong principal, or trusts attacker-controllable state. This allows a malicious HTML page to induce a privileged filesystem operation that should have been denied — and because the operation crosses a trust boundary meant to enforce the sandbox, the downstream effect is code execution outside the sandbox.

The attack surface is the browser's rendering and file-handling pipeline: any HTML page rendered by a vulnerable Chrome instance can attempt to trigger the misauthorized code path. No privileges are required on the attacker's side (PR:N), and no credentials are needed — only that the victim render the crafted page (UI:R). The Scope value of Changed (S:C) reflects that the vulnerability's impact crosses a security boundary: exploitation in the renderer compromises the broader system via the sandbox escape.

Several important technical facts are constrained by the available data:

• The NVD description explicitly attributes the arbitrary code execution to the "outside the sandbox" context, confirming this is a sandbox escape class vulnerability rather than a lower-severity logic flaw.
• The specific function, class, or API within the FileSystem component that contains the authorization error is NOT disclosed in the provided NVD description or CNA title. Chrome security advisories frequently withhold this level of detail until a majority of users have upgraded, to slow weaponization.
• The exact nature of the social engineering required is not detailed. Given the UI:R vector, the victim must interact — for example, clicking a link, opening a downloaded file, or granting a browser prompt.

Because the description is sparse and does not name internal symbols, paths, or endpoints, this analysis deliberately stays at the vulnerability-class level and does not fabricate implementation specifics. No public PoC, exploit code, or Nuclei template is available at the time of writing, consistent with a recently disclosed, high-severity Chrome issue where technical details are typically withheld.

Exploitation Scenarios

No public PoC is available; the attack chain is inferred from the vulnerability class (incorrect authorization → sandbox escape → RCE).

1. Reconnaissance / target identification. The attacker does not need to identify a specific victim system in advance. Any host running Google Chrome earlier than 155.0.8059.39 is a candidate. Attackers commonly identify targets through browsing-habit profiling, watering-hole placement, or by hosting a malicious page on a domain likely to be visited by the target population. Because the CVSS attack vector is Network with No Privileges Required, the attacker's only precondition is delivering a page to a vulnerable browser.

2. Triggering the authorization flaw. The attacker crafts an HTML page designed to invoke a FileSystem operation through a path where the incorrect authorization check resides. Per the vulnerability class, the page induces the browser to perform a privileged filesystem action that the renderer (or its origin) should not be authorized to perform. The victim must be socially engineered into interacting with the page (UI:R) — e.g., clicking a prompt, link, or element that initiates the malicious filesystem operation. The precise API abuse is not disclosed; at the class level, this is the "unchecked/incorrectly checked authorization on a file operation" pattern.

3. Execution / sandbox escape. Successful triggering causes the misauthorized operation to proceed, resulting in execution of attacker-influenced code outside the Chrome sandbox. The Scope:Changed metric captures this boundary crossing. The attacker's code then runs with the privileges of the browser process on the host, not merely within the renderer sandbox.

4. Post-exploitation impact. Once outside the sandbox, an attacker would typically deploy follow-on payloads (RAT, credential stealer, ransomware loader) with user-level system access, pivot to harvest browser-stored credentials and session cookies, and establish persistence. Confidentiality, Integrity, and Availability are all rated High, reflecting potential full compromise of the endpoint's data and operation.

Notably absent from the source data: any indicator of in-the-wild exploitation, any named threat actor, and any observed campaign. CISA KEV status is NO, meaning no confirmed active exploitation as of the available data.

Impact Assessment

The CVSS v3.1 base score is 9.6 (Critical), reflecting the following subscores:

• Attack Vector (AV:N) — exploitable remotely over a network.
• Attack Complexity (AC:L) — no special conditions beyond victim interaction.
• Privileges Required (PR:N) — attacker needs no prior authentication or privileges.
• User Interaction (UI:R) — requires the victim to render and interact with a crafted HTML page.
• Scope (S:C) — impact crosses a trust boundary (renderer → host), hence the "Changed" scope.
• Confidentiality (C:H), Integrity (I:H), Availability (A:H) — all rated High, consistent with arbitrary code execution outside the sandbox.

Business impact is severe for any environment where end users browse the web on Windows, macOS, or Linux desktops and laptops with Chrome installed. Affected systems are individual endpoints rather than servers, but the blast radius can be large across an enterprise: a single successful phishing or watering-hole page can compromise a workstation with the full privileges of the interactive user, enabling credential theft, lateral movement, data exfiltration, and ransomware staging. Because no privileges are required, an attacker can scale campaigns to many targets simultaneously.

The EPSS score has not yet been assigned by FIRST.org, so no probabilistic exploitation estimate is available. CISA KEV enrollment is NO, indicating no confirmed exploitation in the wild as of this writing. There is no evidence of a public PoC, exploit code, or Nuclei template, which somewhat tempers near-term risk — but it does not negate the critical severity of the underlying flaw.

Detection Guidance

Sigma Rule

title: Potential Chrome FileSystem Sandbox Escape Exploitation (CVE-2026-106329)
id: 8f7c2b1a-4d3e-4c9a-9b1e-7a2f6c0d5e11
status: experimental
description: Detects suspicious Chrome child process behavior consistent with a renderer escaping the sandbox via incorrect authorization in the FileSystem component.
references:
  - https://nvd.nist.gov/vuln/detail/CVE-2026-106329
author: valtersit.com Threat Intelligence
date: 2026/01/01
tags:
  - attack.execution
  - attack.t1203
  - cve.2026-106329
logsource:
  category: process_creation
  product: windows
detection:
  selection_parent:
    ParentImage|endswith: '\chrome.exe'
  selection_suspicious_child:
    Image|endswith:
      - '\cmd.exe'
      - '\powershell.exe'
      - '\wscript.exe'
      - '\cscript.exe'
      - '\mshta.exe'
      - '\rundll32.exe'
      - '\regsvr32.exe'
  filter_legit:
    CommandLine|contains:
      - '--type=renderer'
      - '--utility-sub-type'
  condition: selection_parent and selection_suspicious_child and not filter_legit
fields:
  - Image
  - ParentImage
  - CommandLine
  - User
falsepositives:
  - Legitimate browser extensions launching helper processes
  - Enterprise software hooking into Chrome (e.g., DLP, password managers)
level: high

YARA Rule

rule CVE_2026_106329_Chrome_FileSystem_AuthBypass
{
    meta:
        description = "Detects HTML/JS exploitation attempts targeting incorrect authorization in Chrome FileSystem (CVE-2026-106329). Class-level heuristic."
        author = "valtersit.com Threat Intelligence"
        reference = "https://nvd.nist.gov/vuln/detail/CVE-2026-106329"
        date = "2026-01-01"
        severity = "critical"

    strings:
        $fs1 = "requestFileSystem" ascii nocase
        $fs2 = "webkitRequestFileSystem" ascii nocase
        $fs3 = "resolveLocalFileSystemURL" ascii nocase
        $fs4 = "FileSystemDirectoryHandle" ascii nocase
        $fs5 = "FileSystemFileHandle" ascii nocase
        $escape1 = "sandbox" ascii nocase
        $escape2 = "escape" ascii nocase
        $obf1 = "eval(" ascii nocase
        $obf2 = "Function(" ascii nocase
        $obf3 = "atob(" ascii nocase
        $obf4 = "fromCharCode" ascii nocase

    condition:
        uint32(0) != 0x04034b50 and
        (2 of ($fs*)) and
        (1 of ($escape*) or 2 of ($obf*))
}

Suricata IDS Rule

alert http any any -> any any (msg:"CVE-2026-106329 Potential Chrome FileSystem Authorization Bypass Exploit Attempt"; flow:established,to_client; http.response_body; content:"FileSystem"; nocase; content:"requestFileSystem"; nocase; distance:0; within:2048; pcre:"/(eval|Function)\s*\(/i"; classtype:attempted-user; reference:cve,2026-106329; reference:url,nvd.nist.gov/vuln/detail/CVE-2026-106329; metadata:affected_product Google_Chrome, deployment Perimeter, confidence Low; sid:2026106329; rev:1;)

Elastic Detection

process where event.type == "start" and
  process.parent.name == "chrome.exe" and
  process.name in ("cmd.exe", "powershell.exe", "wscript.exe", "cscript.exe", "mshta.exe", "rundll32.exe", "regsvr32.exe") and
  not process.command_line : "(--type=renderer or --type=utility or --utility-sub-type)"

Splunk SPL Query

index=endpoint (source="WinEventLog:Security" OR source="XmlWinEventLog:Microsoft-Windows-Sysmon/Operational")
(EventCode=1 OR EventID=1)
| search ParentImage="*\\chrome.exe" AND Image IN ("*\\cmd.exe","*\\powershell.exe","*\\wscript.exe","*\\cscript.exe","*\\mshta.exe","*\\rundll32.exe","*\\regsvr32.exe")
| where NOT match(CommandLine, "--type=(renderer|utility)")
| stats count min(_time) as first_seen max(_time) as last_seen values(CommandLine) as cmdline by host, User, ParentImage, Image
| where count > 0
| sort - count

Wazuh OSSEC Rule

<group name="windows,sysmon,cve-2026-106329,">
  <rule id="106329" level="12">
    <if_sid>61603</if_sid>
    <field name="win.eventdata.parentImage">chrome\.exe$</field>
    <field name="win.eventdata.image">(cmd|powershell|wscript|cscript|mshta|rundll32|regsvr32)\.exe$</field>
    <description>CVE-2026-106329: Suspicious child process spawned by Chrome, possible FileSystem sandbox escape exploitation</description>
    <mitre>
      <id>T1203</id>
    </mitre>
    <group>cve_2026_106329,</group>
  </rule>
</group>

Mitigation & Patch Guidance

Patch confidence is 0/4 (UNPATCHED) — no patch evidence was retrieved from any tracked source. The NVD description references version 155.0.8059.39 as the fixed release, but no vendor advisory, changelog, or OSV entry confirming the fix was available in the data provided. Treat all affected versions as unpatched until a vendor-confirmed build is deployed and verified.

1. Immediate mitigations:

• Update Google Chrome to 155.0.8059.39 or later as soon as it is available and verified from Google's official release channel. Prioritize this for users who routinely browse untrusted or external web content.
• Where immediate upgrade is not possible, restrict browsing of untrusted content through Chrome using network-level controls, and deploy enterprise allowlist policies for browser navigation where feasible.
• Enforce strict separation of browsing activity from privileged operations. Require privileged administrative tasks to occur on hardened, non-browsing workstations.
• Ensure endpoint detection and response (EDR) coverage is active and alerting on Chrome-spawned child processes, and that browser process telemetry is collected.
• Disable or restrict unnecessary browser extensions and file-download prompts that facilitate the social-engineering stage of the attack.
• Educate users on social engineering lures that require interaction with web content, since UI:R is a hard requirement for exploitation.

2. Patch / upgrade steps:

• Track Google's official Chrome Releases blog and stable channel for the 155.0.8059.39 build.
• Validate the build hash/signature against Google's published binaries before enterprise-wide deployment.
• Use enterprise browser management (Chrome Browser Cloud Management or equivalent) to enforce a minimum version policy and auto-update.
• Confirm the update has applied on all managed endpoints, including remote and BYOD devices accessing corporate resources.

3. Long-term hardening:

• Maintain a rapid browser patch SLA (e.g., critical security updates within 72 hours) given Chrome's role as a primary attack surface.
• Deploy exploit mitigations at the OS level (e.g., Windows Defender Exploit Guard, hardened browser sandboxing).
• Segment networks and apply least privilege so that a compromised endpoint cannot easily pivot to high-value assets.
• Continuously monitor for adversary use of browser-based initial access vectors and integrate CVE-specific detections into the SOC.

References

• NVD — CVE-2026-106329: https://nvd.nist.gov/vuln/detail/CVE-2026-106329
• CWE-863: Incorrect Authorization — https://cwe.mitre.org/data/definitions/863.html
• Google Chrome Releases (Stable Channel updates): https://chromereleases.googleblog.com/
• MITRE ATT&CK T1203 — Exploitation for Client Execution: https://attack.mitre.org/techniques/T1203/
• CISA Known Exploited Vulnerabilities Catalog (status: not listed): https://www.cisa.gov/known-exploited-vulnerabilities-catalog

No known threat actor campaigns or exploit-in-the-wild evidence were identified in the available data. CISA KEV enrollment is NO. No public PoC, exploit code, or Nuclei template exists at the time of writing.