1Password has unveiled a new feature named 1Password for Agents, allowing AI assistants to log into user accounts without ever viewing the password. The service fills in credentials after the user approves each login request.
AI agents, which perform tasks like booking flights or managing subscriptions, often need website and app login details. Handing a password to an AI poses a security risk because copies may end up in uncontrolled locations. 1Password says its tool keeps passwords hidden from the AI at all times.
When an AI agent wants to log in, it first sends a unique digital key to 1Password to prove its identity. The user then receives an alert and can either approve or deny the attempt. After approval, 1Password enters the password itself, ensuring the AI never sees it. The system can operate across multiple sites in a single session.
The feature currently works with several major AI assistants, including Claude Duet from Anthropic, ChatGPT Work and Dots from OpenAI, Grok Bot, Muse from Meta, and agents from a startup called Town. 1Password also says it will add more agents later.
"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 a blog post. "1Password for Agents gives people a trusted way to delegate tasks to personal AI agents while keeping control of the passwords, secrets, and credentials those tasks require."
The launch arrives as concerns grow over AI agents acting on users' behalf. Axios reported the same day that U.S. banks now detect automated activity in customer accounts, though most cannot yet determine if the user authorized it. A Chase spokesperson warned customers not to share usernames or passwords with anyone.
1Password CTO Nancy Wang said the industry needs a new security model designed for AI agents, not just humans. The company protects more than 1.5 billion passwords and serves over 180,000 businesses.
