WebMCP vulnerability allows JS to weaponize browser LLMs

By PulseAugur Editorial · [1 sources] · 2026-05-27 00:01

A new security vulnerability, webMCP, allows malicious JavaScript to register tools with a browser's LLM, bypassing traditional XSS defenses. Researchers demonstrated a complete attack chain, from injection to data exfiltration, within five minutes. While defenses are available, they require more than basic implementation to be effective against this threat. AI

IMPACT This vulnerability highlights a new attack vector targeting browser-based LLMs, potentially impacting the security of AI-integrated web applications.

RANK_REASON The cluster describes a newly identified security vulnerability and its potential impact, which falls under research and safety. [lever_c_demoted from research: ic=1 ai=1.0]

Read on Mastodon — fosstodon.org →

OWASP
webMCP

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

COVERAGE [1]

Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] · 2026-05-27 00:01

OWASP rates XSS as high frequency, low impact. That was before webMCP. Any JavaScript on a page can register tools with the browser's LLM — including injected p

OWASP rates XSS as high frequency, low impact. That was before webMCP. Any JavaScript on a page can register tools with the browser's LLM — including injected payloads. The model trusts both. We validated a full kill chain from injection to C2 exfiltration in five minutes. Defens…

LINKS blog.arachnovato.com/script-injection-mak…

COVERAGE [1]

OWASP rates XSS as high frequency, low impact. That was before webMCP. Any JavaScript on a page can register tools with the browser's LLM — including injected p

RELATED ENTITIES

RELATED TOPICS