Meta launches personal agent Muse with subagent swarms, Sentinel safety layer and agentic purchase protections
jeff_weinstein · x · 2026-09-09
Meta's Superintelligence Labs launched Muse, a personal agent that runs in the background, launches swarms of subagents, builds its own tools, and edits itself — with a 20-minute post detailing its safety engineering.
Highlights:
- Training focused on zero-shot CLI tool calling and skills, long context, long-trajectory instruction following with inherent prompt-injection awareness, and multi-agent coordination.
- Security assumes the agent is under attack: an isolated cell environment, no real credentials, and every external interaction passes through a Sentinel the agent can't override.
- Hardened via dogfooding, agentic red teaming, and adversarial testing by researchers in a private bug bounty program.
- Paired with Link: the Agent Wallet is natively integrated with embedded approvals for human-in-the-loop spending, plus first-of-their-kind purchase protections for agentic shopping (loss, theft, damage, price-drop watching, refund guarantee).
Meta admits Muse will still make mistakes but expects the safety systems to make them rarer and less damaging.
Related event: Meta Launches Muse, an Always-On Personal AI Agent(68 posts)→
More from coding & agent
- Grok bot makes its first real purchase, spots hidden prompt injection at checkout and ignores it — msg · 2026-10-27
- Coding agents now make 60% of Vercel deployments, up from 3% in January — SuB8u · 2026-10-27
- Delta, Chase and Amazon are blocking AI agents — and the blockers may lose, predicts 6-stage war — alvelda · 2026-10-27
- Instinct launches Link balance to let agents spend within set budgets — jeff_weinstein · 2026-10-27
- Production lesson: a fluent AI translation can still be a broken document — WirelessLife · 2026-10-27
- One-prompt test: Grok Bot nails a premium DTC supplement landing hero — azed_ai · 2026-10-27