CP
cyberpings
VulnerabilitiesCRITICAL

Claude Vulnerability - Silent Prompt Injection Attacks Exposed

CSCyber Security News
ClaudeChrome ExtensionZero-Click VulnerabilityArkose LabsXSS
🎯

Basically, a flaw let hackers control an AI tool without users knowing.

Quick Summary

A serious vulnerability in the Claude Chrome Extension exposed millions to silent attacks. This flaw allowed hackers to access sensitive data without user interaction. Thankfully, the issue has been patched, but users must ensure they're updated to stay safe.

The Flaw

A critical zero-click vulnerability was discovered in Anthropic’s Claude Chrome Extension, impacting over 3 million users. This flaw allowed malicious websites to hijack the AI assistant without any user interaction. The vulnerability involved two main issues: an overly permissive origin allowlist and a flaw in a third-party CAPTCHA component.

The first issue was that the extension's messaging API accepted messages from any subdomain of *.claude.ai, which was too broad. The second flaw was linked to Arkose Labs, which provided CAPTCHA verification. A version of their CAPTCHA component had a DOM-based XSS vulnerability, allowing attackers to inject malicious scripts through a hidden iframe on a malicious webpage.

What's at Risk

This vulnerability could have enabled attackers to perform various malicious actions. They could steal Gmail access tokens, read files from Google Drive, and even send emails without the user's knowledge. The attack could be executed silently, meaning users wouldn't see any prompts or alerts indicating that their assistant was compromised. The implications of this vulnerability are severe, as it exposes sensitive personal information and undermines trust in AI tools.

Patch Status

The vulnerability was responsibly disclosed to Anthropic on December 26, 2025. The company confirmed the issue within 24 hours and deployed a fix by January 15, 2026. The patch replaced the wildcard allowlist with a strict origin check, ensuring that only requests from https://claude.ai would be accepted. The XSS vulnerability in the Arkose Labs component was reported separately and patched by February 19, 2026.

Immediate Actions

Users are urged to verify that their Claude Chrome Extension is updated to version 1.0.41 or higher. This can be done by navigating to chrome://extensions. This incident highlights a systemic risk in AI browser agents, where the security perimeter is only as strong as the weakest trusted origin. As AI assistants gain deeper access to browsers, the potential for exploitation increases, making it crucial for users to stay vigilant and keep their software updated.

🔒 Pro insight: This incident underscores the critical need for stringent origin validation in browser extensions to prevent exploitation of trusted components.

Original article from

Cyber Security News · Guru Baran

Read Full Article

Share

Related Pings

HIGHVulnerabilities

NVIDIA Vulnerabilities - Critical RCE and DoS Risks Uncovered

NVIDIA has released critical updates to fix severe vulnerabilities in its AI frameworks. Organizations must patch these flaws to prevent potential remote code execution and denial-of-service attacks. Protect your systems now to avoid significant disruptions and data breaches.

Cyber Security News·
HIGHVulnerabilities

Langflow Vulnerability - CISA Adds Critical Flaw Alert

CISA has identified a critical flaw in Langflow, enabling remote code execution without authentication. This affects many organizations using the tool. Immediate updates are necessary to mitigate risks and protect systems.

Security Affairs·
HIGHVulnerabilities

HPE Telco Service Orchestrator - Security Advisory Released

HPE has issued a security advisory for vulnerabilities in its Telco Service Orchestrator software. Users of versions before v5.5.1 must update to avoid security risks. This is crucial for maintaining system integrity and preventing potential breaches.

Canadian Cyber Centre Alerts·
HIGHVulnerabilities

Spring Security Advisory - Critical Vulnerabilities Identified

Spring has issued a security advisory for vulnerabilities in Spring Cloud Config and Spring AI. Users must update their software to prevent unauthorized access and remote code execution. Ignoring these updates could lead to serious security risks. Stay secure by applying the necessary patches.

Canadian Cyber Centre Alerts·
HIGHVulnerabilities

Grafana Vulnerabilities - Critical Security Advisory Issued

Grafana has issued a critical security advisory for older versions. Users must update to avoid serious vulnerabilities. Acting now is essential for safeguarding data integrity.

Canadian Cyber Centre Alerts·
HIGHVulnerabilities

ABB Security Advisory - Addressing Camera Connect Vulnerabilities

ABB issued a security advisory for vulnerabilities in their camera software. Users must upgrade to avoid potential exploitation. This is crucial for maintaining security in connected systems.

Canadian Cyber Centre Alerts·