Inappropriate implementation in Glic in Google Chrome prior to 150

This is an *inappropriate implementation* bug (CWE-451, UI misrepresentation) inside Glic, the Chromium component that hosts the Gemini AI side panel. A remote attacker who convinces a user to perform a specific interaction sequence can render the Glic surface in a way that misleads the user about the origin, trust boundary, or content of what they're seeing. Affected versions are all Chrome stable builds prior to 150.0.7871.47. There is no memory corruption, no sandbox escape, no arbitrary code execution — the impact is confined to a UI spoofing / information-disclosure primitive that the attacker still has to socially engineer around.

Vendor severity of MEDIUM (5.3) with vector AV:N/AC:H/PR:N/UI:R/S:U/C:H/I:N/A:N is honest — arguably slightly generous. The C:H rating assumes the spoof reliably exfiltrates something sensitive (e.g., tricks a user into pasting secrets into a fake-looking Gemini prompt), but AC:H + UI:R means the attacker needs both a specific victim behavior AND a low-reliability trigger. Real-world exploitability against a fleet is closer to LOW.

Why this verdict

• No code execution path. CWE-451 spoofing bugs cap out at social-engineering uplift; there is no memory corruption, type confusion, or sandbox weakness described.
• AC:H + UI:R stacked. The attacker needs a fragile trigger AND a cooperative user AND a valuable secret to paste. Real yield against a hardened workforce is minimal.
• Narrow reachable population. Glic is region- and policy-gated; enterprise fleets with GenAiDefaultSettings=Disabled are not exposed at all.
• Chrome auto-update dominates. The typical enterprise Chrome fleet reaches >90% patched within 7 days without any admin action — the mitigation window is largely self-closing.
• Role multiplier: workstation. Chrome runs on user endpoints, not identity/hypervisor/CI infrastructure. There is no fleet-scale blast radius from a browser UI spoof — the deployment-role floor stays at LOW/MEDIUM.
• Not on KEV, no public PoC, no exploitation reports as of assessment date.

Why not higher?

MEDIUM would be defensible if there were a credible PoC combining the Glic spoof with a follow-on primitive (e.g., prompt-injection to Gemini's authenticated context). Absent that chain, the vector's C:H claim is theoretical — the bug is a phishing accessory, not a browser compromise. Nothing here justifies HIGH.

Why not lower?

IGNORE is wrong because the CVE is real, the affected code ships enabled to many users, and CWE-451 has been chained into credential-theft campaigns before. It deserves a tracked patch — just not a fire drill.

1. Force Chrome auto-update and verify fleet ≥ 150.0.7871.47 — Chrome's built-in updater is the primary control. Push ChromeCleanupEnabled + RelaunchNotification=Required via GPO/MDM with a 7-day relaunch deadline — well inside the noisgate 365-day remediation window for a LOW verdict.
2. Disable Glic / Gemini side panel in managed environments where the feature isn't sanctioned — Set GenAiDefaultSettings=1 (Do not allow) via Chrome Enterprise policy. Eliminates the attack surface entirely on managed endpoints. No mitigation SLA at LOW — treat as backlog hygiene, deploy on next policy cycle.
3. Enable Enhanced Safe Browsing fleet-wide — Blocks the delivery-page phase of the chain for the majority of drive-by lures. Set SafeBrowsingProtectionLevel=2 via policy.
4. Awareness note to users, not a training campaign — One-line advisory in the security channel: *'Don't paste credentials or secrets into browser AI side panels, ever.'* Full CBT is overkill for a LOW.

Run on any Windows workstation to check the installed Chrome version against the fix. Invoke from an admin or user PowerShell prompt: .\Check-CVE-2026-14153.ps1. No elevation required; reads registry only.

#Requires -Version 5.0
# Check-CVE-2026-14153.ps1
# Reports VULNERABLE / PATCHED / UNKNOWN for Chrome Glic UI-spoof bug
# Fixed in Chrome 150.0.7871.47

$ErrorActionPreference = 'Stop'
$FixedVersion = [Version]'150.0.7871.47'

function Get-ChromeVersion {
    $paths = @(
        'HKLM:\SOFTWARE\Google\Update\Clients\{8A69D345-D564-463C-AFF1-A69D9E530F96}',
        'HKLM:\SOFTWARE\WOW6432Node\Google\Update\Clients\{8A69D345-D564-463C-AFF1-A69D9E530F96}',
        'HKCU:\SOFTWARE\Google\Update\Clients\{8A69D345-D564-463C-AFF1-A69D9E530F96}'
    )
    foreach ($p in $paths) {
        try {
            $v = (Get-ItemProperty -Path $p -ErrorAction Stop).pv
            if ($v) { return [Version]$v }
        } catch {}
    }
    # Fallback: binary FileVersion
    $exe = @(
        "$env:ProgramFiles\Google\Chrome\Application\chrome.exe",
        "${env:ProgramFiles(x86)}\Google\Chrome\Application\chrome.exe",
        "$env:LOCALAPPDATA\Google\Chrome\Application\chrome.exe"
    ) | Where-Object { Test-Path $_ } | Select-Object -First 1
    if ($exe) {
        return [Version]((Get-Item $exe).VersionInfo.ProductVersion)
    }
    return $null
}

$installed = Get-ChromeVersion
if (-not $installed) {
    Write-Output 'UNKNOWN - Chrome not detected'
    exit 2
}

if ($installed -lt $FixedVersion) {
    Write-Output ("VULNERABLE - Chrome {0} < {1}" -f $installed, $FixedVersion)
    exit 1
} else {
    Write-Output ("PATCHED - Chrome {0} >= {1}" -f $installed, $FixedVersion)
    exit 0
}

Sources

1. Chrome Releases blog (Stable channel updates)
2. Chrome Enterprise policy list — GenAiDefaultSettings
3. CWE-451: User Interface (UI) Misrepresentation of Critical Information
4. CISA Known Exploited Vulnerabilities Catalog
5. FIRST EPSS model documentation
6. Chromium bug tracker (public entries)
7. Bleeping Computer — Chrome zero-day coverage
8. The Hacker News — Chrome Gemini panel vulnerability