Skip to content
NewA calmer, faster ZeroHour

Muse share links expose private chats through DNS-visible names

Original title: A Design Flaw in Share Links Exposes Users’ Personal Conversations With Meta’s Muse

WIRED · Securityby Dan Goodin, Ars Technica ()4 min read1 · 1 read
Part of a story covered by 6 sourcesMeta restores Muse sharing after replacing discoverable linksMerged summary & timeline
AI summary · x-ai/grok-4.7Summaries by Vani AI

Claroc enumerated 6,800 links from passive DNS; Meta has paused sharing while it reviews the report.

Claroc said Meta’s Muse put conversation titles in subdomains of cf.metaaiusercontent.com, so passive DNS recorded share links after a recipient opened them. Researchers enumerated 6,800 links with one query and, after opening a few hundred, found material such as executive-search notes, a revenue review, a loan application, and a partial bank-account number. Meta told Ars it paused sharing while reviewing the October 8 bug-bounty report. Cofounder Asaf Aprozper said outreach on LinkedIn also got no reply.

Unconfirmed: The contents of opened links are Claroc’s account; Meta has not publicly confirmed those specific exposures.

Key facts
  • Example hosts include titles such as q3-revenue-review plus a long token.
  • Claroc calls the discoverable pages GhostShares.
  • Findings were sent through Meta’s Muse bug bounty on October 8.
  • Meta said people with the link could open a shared artifact page.
  • A separate Muse macOS account-takeover issue reported by Patrick Wardle is mentioned only as background.
Security assessment · AI, from the articlePublic proof of concept

Impact: Private Muse chats and personal or business details exposed

ProductAffected
Meta Muse—
What to do
  1. 1.Treat previously created Muse share links as discoverable while Meta reviews the feature.
Explain this AI news with Vani AIWhat was announced, what is new and why it matters, in a new chat

From the original article

4 min read at wired.com

The company is suspending the feature after security researchers discovered how easy it was to access sensitive information shared with the AI assistant.

WIRED · Security

By Dan Goodin, Ars Technica · wired.com

Read the full article at WIRED · Security

A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.