What happened
Kimi's feature for previewing and publicly sharing LLM-generated HTML artifacts fails to properly sanitize content before rendering, allowing injected script to execute in the context of any user who views the shared/previewed artifact.
Why it matters
Kimi is a widely-used consumer LLM chatbot with a large user base; XSS in a feature designed to let users share AI-generated content publicly creates a straightforward vector for session hijacking or credential theft against anyone who opens a malicious shared link.
Attack vector
A cross-site scripting vulnerability in Kimi's HTML artifact Preview rendering and public Share view component allows a remote attacker to execute arbitrary JavaScript in a victim's browser session, likely via a crafted shared artifact or chat output rendered as HTML.
Affected systems
Moonshot AI Kimi (version as of 2026-07-18)
Mitigation
Vendor patch pending/applied per Kimi's own platform (kimi.com); users should treat shared artifact links from untrusted sources with caution until confirmed patched.