Meta's Muse agent can touch your email and money — and admits it's prompt-injectable, $130k bounty offered
JanJanJaJa · reddit · 2026-09-10
Meta's new agent Muse can read email, book things and pay — and its own safety docs admit it "remains susceptible to adaptive jailbreaks and prompt injection," with bounties up to $130k (of a $300k pool).
The post draws parallels to Anthropic's recent Gmail access for Claude, where hidden white-on-white text or invisible Unicode in emails was shown to steer the agent. The industry's fixes — sandboxes, watcher agents (Meta's "Sentinel"), intercepting OTPs — are dismissed as stronger cages around the same flawed base: plain-text email built for human eyes repurposed as instructions for AI. The author, who builds AI-native email (Atomic Mail Agentic), asks why nobody is rethinking the protocol itself.
Related event: Meta Offers Up to $300K Bug Bounty for Muse Agent(4 posts)→
More from coding & agent
- AIE Code Summit 2026 opens applications: 800 seats, keynotes streamed free — trq212 · 2026-10-27
- TRL adds Score Centering, a simple fix for RLHF training instability — lvwerra · 2026-10-27
- Precompute golden answers and match at runtime for exact wording control — colinmcnamara · 2026-10-27
- Dev argues there's still alpha in working with your coding agent interactively — alex_frantic · 2026-10-27
- Researchers launch AMA on patterns for safer AI agents used by citizen developers — Loose_Rip359 · 2026-10-27
- Gradio Adds In-Chat Human-in-the-Loop So ML Intern Pauses Before Training — Gradio · 2026-10-27