- Both run dedicated cloud computers with their own browsers and connected-app access, continuing tasks after the user leaves.
- Muse sends email, books travel, completes forms, and makes purchases; it is free for most uses with paid tiers for heavier use.
- Dots conduct background research through connected apps in read-only mode — those tools cannot send messages, change app content, or control a browser or computer.
- Users can create rules for Dots; individual memories cannot currently be viewed, deleted, or directly modified.
- Muse requests approval before actions such as sending email or making a purchase.
- Dots require the user to take over for password changes and money transfers.
- Deleting data or installing software may require approval; read-only background tools are the default for research.
- Muse maintains an audit trail covering completed and planned actions.
- Conversations, tool calls, and subagent exchanges may be sanitized and used to train future models by default, with an opt-out in Muse settings.
- Muse interactions and VM data are not shared directly with Meta's ad systems, though agent browsing can influence advertising indirectly.
- Meta's security program pays up to $300,000 for valid bug reports including prompt injection; a Confidential VM encrypting the machine with a user-held key is planned for later in 2026.
- OpenAI says human review may occur in limited situations, including safety cases, even when model improvement is disabled.
Useful?