Direct answer
Direct answer: 1Password has launched 1Password for Agents, a tool that lets AI agents log into your accounts on your behalf without the agent ever seeing your password. When the agent attempts a login, you get an alert and approve it personally; 1Password then fills in the credentials for it. It lands the same week US banks told Axios they are detecting unexplained agent logins inside customer accounts.
What happened
Password manager 1Password unveiled the new platform on Thursday 22 October, responding to a growing problem: people are handing over passwords so AI agents can complete tasks, and there has been no safe way to do it.
The mechanism is straightforward. The agent, the user and the password manager communicate with each other: a unique digital key confirms the agent’s identity to 1Password, the user approves the login attempt from an alert, and 1Password signs the agent in itself — filling in the credentials so the agent never holds them.
At launch it supports Anthropic’s Claude Duet, OpenAI’s Dots and ChatGPT Work, Grok Bot, Meta’s Muse, and agents from the startup Town, with more integrations promised.
“Handing over a password, secret, or credential is a common shortcut to an agent, but can leave people with sensitive information in places they can no longer control as a copy gets saved for the next time,” the company said in its launch blog post.
The timing is pointed. On the same day, Axios reported that US banks have started detecting agentic activity inside user accounts — not necessarily fraud, but consumers letting automated services act on their behalf, including on financial matters. The problem, bankers say, is that most banks cannot tell whether the agent activity is tied to a real user request. A Chase spokesperson told Axios customers “should not give their usernames and passwords to anyone.”
What it means for UK small businesses
- Safer agentic commerce. If your customers use AI agents to buy from you or manage bookings, credential-delegated logins mean fewer shared passwords floating around.
- Lower breach risk. Employees already paste passwords into AI tools; a formal approval flow keeps credentials inside the password manager instead of inside chatbot transcripts.
- Trust signal for your own agents. If you run an AI agent that logs into third-party services for customers, supporting approved credential delegation is a strong answer to security questions.
- Relevance to UK fraud rules. Under UK authorised push-payment rules, clear user approval trails matter — tools that log who approved what are useful evidence.
For context on the wider risk landscape, see our piece on rogue AI agents breaking into live systems and the new $42M AI security funding round.
Honest limits / what to watch
- This only covers the login step. It does not stop an agent from doing something unwise once inside the account.
- It depends on the agent supporting the protocol — most agents today still expect you to paste credentials.
- Banks and services can still flag agent logins as suspicious, which may lock accounts.
- Availability and pricing details are limited; confirm it fits your 1Password plan before relying on it.
FAQs
How does 1Password for Agents work?
When an agent tries to log in on your behalf, a unique digital key is sent to 1Password to confirm the agent’s identity. You get an alert and approve the request personally. 1Password then fills in the credentials for the agent, which never sees the password itself.
Which AI agents are supported?
At launch it works with Claude Duet (Anthropic), OpenAI’s Dots and ChatGPT Work, Grok Bot, Meta’s Muse, and agents from Town, an AI assistant startup. 1Password says it is working with more companies to expand support.
Why not just give my AI agent my password?
Because a password copied into an agent gets saved and stored in places you no longer control. A bank spokesperson told Axios customers should not give their usernames and passwords to anyone — and US banks are already detecting agent logins they cannot attribute to a real request.
Is this a business tool or for consumers?
It is aimed at individuals delegating personal tasks to AI agents — shopping, bookings, account admin — where logging in is required. Businesses using agents that touch customer or finance accounts should look at 1Password’s enterprise agent offerings instead.
When is it available?
1Password launched the platform on 22 October 2026. Check 1Password’s announcement for availability in your region and plan.
