1Password for Agents: your AI agent can sign in to your accounts without ever seeing the password

• Who: 1Password, announcing from Toronto on October 22, 2026
• What: 1Password for Agents — a sign-in platform that lets AI agents log in without receiving or storing passwords
• Launch partners: Meta's Muse and SpaceXSI's Grok Bot go live first; Anthropic's Claude Duet, OpenAI's ChatGPT Work and dots, and Town's Townie agents are listed as coming soon
• Also building: Browserbase, Browser Use, Dana, Instinct, KERNEL, Perplexity, Ramp, Tomo and Wajo — plus a waitlist for agents not yet on the list
• Price: included at no extra cost in 1Password Individual and Family plans on 1Password.com; accounts on 1Password.eu and 1Password.ca are not supported
• Scale: 1Password says its vaults protect more than 1.6 billion credentials and secrets, used by over 200,000 businesses

1Password announced 1Password for Agents on Thursday, October 22, 2026 — a platform that lets personal AI agents sign in to your accounts without ever seeing or storing your password. The Toronto company describes it as a trust layer for the personal-agent era: the agent asks, you approve on your own device, and 1Password fills the credential into the login form itself.

The pitch is aimed at the last big friction in delegating work to an AI agent. Booking a flight, reordering groceries, or reconciling an expense report all sit behind logins, and until now the options were bad: type the password in yourself, hand over the browser, or paste the credential into a chat window — where a copy lives outside your control from then on.

How a sign-in works

The flow has four steps, per 1Password's announcement:

1. The agent identifies itself. When it reaches a sign-in form, the agent presents a unique digital key to 1Password, proving the request came from, say, Grok Bot rather than an arbitrary script pretending to be it.

2. You get the request. You see which agent is asking, which login it wants, and why — and you approve or deny on your own device. John Torrey, 1Password's chief business officer, told PCWorld that the agent cannot approve its own access: "The user is involved in every decision."

3. 1Password fills the form. Instead of handing the secret to the agent, 1Password enters the username, password, and any one-time code directly into the site's sign-in form through a channel it controls.

4. The agent keeps working. From the agent's side it only learns that access was granted. It can continue the task while the actual password stays inside your vault.

Approvals are per task and per session — access does not carry over to later work, and agents can only use logins you have explicitly approved, not the rest of your vault. 1Password says you can disconnect an agent integration at any time.

Who is in, and who is not

Six agent products are named as launch partners: Meta's Muse, SpaceXSI's Grok Bot, Anthropic's Claude Duet, OpenAI's ChatGPT Work and dots, and the startup Town's Townie agents. Not all are live on day one: Muse and Grok Bot are rolling out over the coming days, while Duet, ChatGPT Work, dots, and Townie are listed as coming soon. Anthropic's Ami Vora said 1Password will "soon" complete sign-ins for Claude Duet, without giving a date.

Nine more companies are building integrations for later — Browserbase, Browser Use, Dana, Instinct, KERNEL, Perplexity, Ramp, Tomo, and Wajo — and agents not on the list can be requested through a waitlist. Notably absent from the launch roster: Google's Gemini Spark and Microsoft's Scout.

The feature is included at no extra cost in 1Password's Individual and Family plans on 1Password.com, where new users get a 14-day trial. Accounts hosted on 1Password.eu and 1Password.ca are excluded at launch, and 1Password says availability and safeguards vary by integration.

What this does not cover

The honest fine print matters here. Keeping the password hidden does not make the delegation harmless: once the sign-in completes, the agent is operating inside a logged-in session and can do whatever that session allows — place an order, change a setting, send a message. The password is safe; the account itself is in the agent's hands for the duration of the task. That is why 1Password's own framing stresses approval per task rather than blanket trust.

There is also a design bet embedded in the launch. Four rival AI companies — Meta, Anthropic, OpenAI, and SpaceXSI — all agreed to route their agents' sign-ins through the same third party. If that holds, users get one consistent approval habit instead of a different credential scheme for every assistant, and the password manager — rather than the agent vendor — becomes the gatekeeper for account access. Browsers and operating systems may want that job for themselves.

1Password for Agents builds on 1Password for Claude, which launched July 16, 2026 and worked only on the Mac inside Chrome. The new platform extends the same no-exposure approach to agents that are not tied to a single desktop browser session. With non-human credential usage reportedly surging on its platform, 1Password is clearly betting that delegated sign-ins become a core password-manager feature rather than a niche add-on.