Claude Chrome Extension 0-Click Vulnerability Enables Silent Prompt Injection Attacks

A critical zero-click prompt-injection vulnerability in Anthropic’s Claude Chrome extension—caused by an overly broad *.claude.ai origin allowlist and a DOM XSS in an Arkose Labs CDN component—could be exploited via a hidden iframe to silently execute attacker-controlled prompts, enabling theft of Google OAuth tokens, access to Gmail/Drive, and exfiltration of chat history for over 3 million users; the issues were responsibly disclosed and patched in Jan–Feb 2026 (users should upgrade to version 1.0.41+).