ShadowPrompt Flaw Let Websites Inject Prompts into Claude Chrome Extension

Summary
Researchers reported that ShadowPrompt chained a permissive origin allowlist and a CAPTCHA-component XSS to inject prompts into Claude’s Chrome extension. Anthropic patched the extension, and Arkose Labs fixed the XSS.
Key points
- The reported flaw let a website inject prompts into Claude’s extension without user interaction.
- The exploit chained an overly broad origin allowlist with a DOM-based XSS in an Arkose Labs CAPTCHA component hosted at a-cdn[.]claude[.]ai.
- Potential impacts included access-token theft, access to conversation history, and actions such as sending emails as the user.
- Anthropic’s extension patch, version 1.0.41, enforced an exact match for claude[.]ai; the article says it was deployed on December 27, 2025.
- Arkose Labs fixed the XSS on February 19, 2026.
Article Details
- Vulnerability Types
- DOM-based cross-site scripting (XSS)
- Overly permissive origin allowlist
- Severity
- Unknown
- Exploitation Status
- unknown
- Exploit Availability
- unknown
- Patch Status
- available
MITRE ATT&CK
People
Vendors
AnthropicResearchers disclosed “ShadowPrompt,” a vulnerability in Anthropic’s Claude Chrome extension that allowed any website to silently inject prompts by chaining an overly permissive (*.claude.ai) origin allowlist with aArkose Labsprompts by chaining an overly permissive (*.claude.ai) origin allowlist with a DOM-based XSS in an Arkose Labs CAPTCHA component.
