JavaScript Injection in AI Model Safety Scanner Browser Automation

ai-scanner is an AI model safety scanner built on NVIDIA garak. From version 1.0.0 to before version 1.4.1, there is a remote code execution vulnerability via JavaScript injection in `BrowserAutomation::PlaywrightService`. This issue has been patched in version 1.4.1.

Casky was already ahead

This CVE exploits attack patterns that Casky's 301matched skills already investigate — long before this vulnerability was disclosed. Claude's reasoning model maps these techniques to MITRE ATT&CK, so practitioners who ran these skills have already seen the threat behaviour in their findings.

Analysis

CVE-2026-41512 is a critical remote code execution vulnerability in ai-scanner's BrowserAutomation::PlaywrightService that allows attackers to inject malicious JavaScript code, bypassing the safety mechanisms of an AI model security tool. This vulnerability affects versions 1.0.0 through 1.4.0 and is particularly dangerous because it compromises a defensive security product—turning a tool designed to detect AI model exploits into an attack vector itself. Organizations using ai-scanner for model safety validation, security researchers, and AI development teams relying on this NVIDIA garak-based tool are at direct risk of having their security posture undermined through JavaScript injection payloads.

Casky's 301 matched skills, powered by Claude AI with extended reasoning capabilities, detect the attack patterns behind this vulnerability by mapping to MITRE ATT&CK techniques TA0001 (Reconnaissance) and TA0043 (Reconnaissance). Practitioners using Casky would identify suspicious JavaScript injection attempts through code analysis skills that examine input sanitization failures in browser automation contexts, spot malicious payload patterns in Playwright service calls, and detect CWE-94 (Improper Control of Generation of Code) indicators. When analyzing logs or code repositories, Casky's skills would surface evidence of unsanitized user input flowing directly into JavaScript execution contexts, flagged payloads attempting to break out of the Playwright sandbox, and behavioral signals indicating post-exploitation command execution—enabling security teams to rapidly identify both vulnerable deployments and active exploitation attempts targeting their AI safety infrastructure.

Live Threat IntelligenceComing soon

Casky Risk Score

—

EPSS Probability

—

Shodan Exposure

—

GreyNoise Signal

—

Composite risk scoring from EPSS, CISA KEV, Shodan, and GreyNoise — 21 security APIs correlated into a single Casky Risk Score. Coming in Casky Pro. Join early access →

301 Casky skills cover this attack pattern

These skills use Claude AI's reasoning model to surface findings in the same attack categories as CVE-2026-41512.

analyzing-apt-group-with-mitre-navigator

threat intelligence · low

Run

analyzing-campaign-attribution-evidence

threat intelligence · low

Run

analyzing-certificate-transparency-for-phishing

threat intelligence · low

Run

MITRE ATT&CK Techniques

TA0001 TA0043

Investigate the attack patterns behind CVE-2026-41512

Casky has 301 skills that investigate the attack patterns behind CVE-2026-41512. Run one and get CVSS-scored findings in 3 minutes.

Run the skill that detects this →

X Instagram LinkedIn