1Password Launches System for AI Agents to Log In Without Seeing Passwords
New feature lets approved AI assistants sign into accounts via 1Password without ever accessing your actual login credentials.
Key points
- Password for Agents allows AI tools to sign in without storing or seeing passwords.
- Users must approve each login request on their personal device.
- Initial partners include OpenAI, Meta, Anthropic, and Town.com.
1Password has launched a new service that lets artificial intelligence agents sign into user accounts without ever accessing the actual passwords. This system keeps credentials secure while allowing automated tools to perform tasks like booking travel or paying bills.
In plain English
Previously, users might paste passwords into chat windows for AI agents to use, leaving copies saved outside their control. The new system avoids this risk entirely. The password never travels to the agent. Instead, the agent uses a unique digital key to identify itself to 1Password when it reaches a login page.
1Password then sends a notification to the user’s device explaining which account is needed and why. The user must approve this request personally. Once approved, 1Password fills in the login information directly into the web page. The agent proceeds with its task, but it never sees or handles the password itself.
The background
Aashish Tripathi, chief product officer at 1Password, stated that agents are starting to do real work on behalf of users. He noted that the old model of handing over credentials and hoping for the best no longer works. An agent performing a specific job should not get permanent access to everything behind an account.
This new platform provides a trust layer for these interactions. It ensures that users remain in control of permissions. The process is similar to two-factor authentication, requiring active approval from the account owner on their own device before any sign-in occurs.
What changes now
Several major companies have integrated with 1Password for Agents at launch. OpenAI connected ChatGPT Work and its always-on Dots agents to the platform. Meta’s Muse agent can request logins through 1Password when the integration is enabled. Town.com Inc. has also signed up for the service.
SpaceXSI’s Grok Bot can now sign in without users pasting passwords, according to head of product Tido Carriero. Anthropic’s Claude Duet agent asks before signing in anywhere. Anthropic Chief Product Officer Ami Vora said 1Password will soon fill in those sign-ins using existing passwords.
Nine other companies are building their own integrations, including Perplexity AI and Browserbase. Canva and Stripe are among more than 200,000 businesses that currently use 1Password for identity security.
Key terms explained
When using AI agents for account access, teams usually verify that credentials are not stored in chat logs or external servers. Check whether the service requires explicit user approval for each login attempt. Ensure that the authentication method keeps the actual password isolated from the agent’s memory. This reduces the risk of accidental data exposure during automated tasks.
What to do and how to stay safe
- Review which AI agents you currently use and whether they require password access.
- Enable multi-factor authentication on all accounts that allow agent integration.
- Monitor login notifications closely and only approve requests you
Step-by-step guide: Stop AI Chatbots Leaking Sensitive Business Data
General security guidance from the Vector Update newsroom. It is not confirmed advice from the organisations named in this story.
