An AI assistant can plan the trip and still get stuck at the airline’s login page. 1Password’s answer is to let it ask for access while you keep control of the password. Announced on 22 October, 1Password for Agents puts an approval step between the assistant and a saved login.

Based on 6 reviewed sources · See sources

Permission at the door

The product FAQ describes a simple sequence: the agent requests a login, you approve or deny it on your own device, and 1Password fills the password or one-time code into the site. The company says the model does not receive the secret, and connecting an agent does not open the whole vault.

Its original announcement below introduces that approach. The important distinction comes after the door opens: the FAQ says an agent may then see information or take actions inside the account, depending on the assistant and the website. Protecting the password does not settle what the assistant is allowed to do next.

From the source · 1Password · @1Password ·

1Password announces a platform for user-approved agent sign-ins without handing the model a password.

View the original post on X ↗

The partner list is not a launch checklist

1Password’s launch blog describes Muse and Grok Bot connections as rolling out over the coming days. Claude Duet, ChatGPT Work, dots and Townie are listed as coming soon. The logos in the announcement therefore cover different stages of availability.

The feature is optional and included at no extra cost for Individual and Family accounts on 1Password.com. Accounts hosted on 1Password.eu and 1Password.ca are not yet supported. For European readers, that is an account-hosting restriction; it should not be mistaken for a blanket ban on people living in Europe.

Some doors still need you

In a direct reply to a question about passkeys, 1Password confirmed on 23 October that the initial release cannot sign in to passkey-only accounts. Passkeys use a cryptographic sign-in rather than a typed password. Support is planned, with no release date given in the reply.

This is a sign-in announcement, not an independent security audit or a test of a completed errand. The useful question for the next hands-on review is whether people can clearly understand, limit and withdraw access after approving a login. A hidden password is a useful boundary; the signed-in account needs boundaries too.

Sources

Written with AI assistance from the sources above. Analysis and practical implications are our interpretation. Our editorial approach.