CVE-2026-9908: Chrome ANGLE Out-of-Bounds Memory Read – Data Leakage Vulnerability

Technical summary

The vulnerability is an out-of-bounds read in ANGLE (Almost Native Graphics Layer Engine), Chrome's WebGL/graphics abstraction layer. CWE-125 classifies this as a read access beyond the bounds of an allocated buffer. During graphics processing or shader compilation, the code fails to validate buffer indices or bounds before reading, allowing an attacker-controlled shader or WebGL command sequence to access adjacent memory. The disclosed data is then exfiltrated via timing side-channels, error messages, or pixel readback operations. This is a confidentiality issue with no impact on integrity or availability.

Business impact

Data leakage from browser process memory can expose customer credentials, authentication tokens, personal information, or intellectual property in use within the browser context. For organizations where employees browse sensitive web applications or handle confidential documents via web interfaces, this increases the risk of credential compromise and lateral movement. The attack surface is broad since it requires only a malicious webpage visit, making it suitable for targeted attacks against specific user populations.

Affected systems

Google Chrome versions prior to 148.0.7778.216 are vulnerable. The issue affects the ANGLE graphics component, which is used on Windows, macOS, and Linux. Any system running an unpatched Chrome build is at risk if the user visits a malicious or compromised website.

Exploitability

Exploitability is moderate to high. The attack requires user interaction (visiting a crafted HTML page), but no special privileges or authentication. The attacker needs only to host the malicious page or compromise an existing site. The technical barrier to crafting a WebGL or shader-based exploit is moderate; it requires knowledge of graphics APIs but not zero-day development. CVSS 6.5 reflects the high confidentiality impact and low attack complexity, tempered by the requirement for user interaction.

Remediation

Update Google Chrome to version 148.0.7778.216 or later. Automatic updates should deploy the patch; verify the installed version in chrome://version/. For managed environments, use Chrome policies to enforce updates and block older versions. In the interim, security controls cannot fully mitigate this remotely exploitable flaw, but blocking WebGL or restricting access to untrusted sites may reduce exposure for high-risk users.

Patch guidance

Deploy Chrome 148.0.7778.216 or later via auto-update or manual installation. Verify rollout completion by checking version strings in deployed endpoints. No workarounds are available; patching is the only reliable remediation. Monitor Chrome update adoption in your environment to ensure timely coverage.

Detection guidance

Detection at the network level is difficult since the attack occurs within JavaScript/WebGL execution and memory access is local to the process. Behavior-based detection could flag unusual WebGL shader compilation patterns or repeated pixel readback operations, but this requires endpoint telemetry and EDR integration. Monitor for suspicious JavaScript targeting WebGL APIs in proxy logs. At the endpoint, EDR tools may detect process memory scanning if the exfiltration method involves abnormal IPC or network transmission of gathered data.

Why prioritize this

Although rated CVSS 6.5 (MEDIUM), this vulnerability merits prioritization because it enables silent, user-interaction-dependent data exfiltration affecting a widely used browser. It is not listed in CISA KEV, but the broad attack surface, ease of delivery, and confidentiality impact make it a suitable candidate for targeted campaigns. Prioritize patching for users accessing sensitive web applications or handling high-value data in the browser.

Risk score, explained

The CVSS 3.1 score of 6.5 reflects a network-accessible, low-complexity attack requiring user interaction and no privileges, with high confidentiality impact but no integrity or availability impact. The lack of KEV listing suggests no known active exploitation as of the publication date, but the attack vector and practical exploitability remain significant in real-world scenarios.

Frequently asked questions