AiGpu

Industry·

Meta's Muse AI Agent Reportedly Leaks Private Data Across Accounts

Internal reports reveal that Meta's rapidly growing AI assistant Muse is mixing up user conversations and exposing sensitive personal information.

Illustration of an AI assistant mixing user data across multiple accounts

Meta's new AI assistant Muse, launched in September, has quickly surpassed ten million downloads, yet internal posts indicate serious privacy flaws.

Employees report that Muse conflates conversations across different accounts, surfacing unrelated childcare requests, career advice, product orders, and even bank balances belonging to other users.

In one instance the agent inserted a message listing exact Capital One and Charles Schwab balances; in another it reminded a user to reply to a Signal message without any granted permission.

Why it matters for GPU / AI infrastructure

The episode highlights the critical need for strict data isolation and access controls in large‑scale AI services, a priority for any cloud provider hosting generative models.

  • aigpu
  • ai gpu
  • ai gpu cloud
  • aigpu dubai
  • meta
  • privacy
  • ai-agent
  • data-leak

By AiGpu Editorial · Editorial rewrite based on public reporting (TechCrunch AI)

← All articles