1Password Lets AI Agents Log In, Without the Password

As AI agents increasingly act on people’s behalf online, 1Password has launched a way for them to log into accounts without ever seeing the user’s actual password, tackling one of the thorniest security problems of the agentic era.

Announced on Thursday, October 22, 2026, the feature, called 1Password for Agents, lets users authorise an AI agent to sign into their personal accounts while the credentials stay firmly under the password manager’s control.

Key Highlights

• 1Password for Agents launched Oct 22, 2026, letting AI agents log in without seeing passwords
  A digital key verifies the agent; the user approves each attempt; 1Password fills credentials itself
  Launch partners include Claude Duet, OpenAI’s Dots/ChatGPT Work, Grok Bot, Meta’s Muse and Town
  Lands as US banks detect unverified agentic activity, with Chase warning against sharing credentials
  The problem it solves is real and growing. Today, the common shortcut is to simply hand an agent your username and password, a practice that scatters sensitive information into places users can no longer control, with copies potentially lingering long after the task is done.

The mechanism is elegantly simple. When an agent tries to log in, a unique digital key is sent to 1Password to verify the agent’s identity.

The user then receives an alert about the login attempt and can explicitly confirm or deny it, keeping a human firmly in the loop.

Only after approval does 1Password sign the agent in by filling the credentials itself, so the agent completes the task without ever seeing the password.

The security model rests on three pillars: 1Password retains custody of credentials, it verifies each agent’s identity, and it demands explicit user approval for every login.

The launch arrives with notable partners, including Anthropic’s Claude Duet, OpenAI’s Dots and ChatGPT Work, Grok Bot, Meta’s Muse and agents from Town, with more to come.

The timing is pointed. The same day, Axios reported that US banks are detecting agentic activity in customer accounts but often cannot tell whether it reflects genuine user requests, prompting Chase to warn customers never to share credentials.

Ultimately, 1Password is betting that the future of agentic AI hinges on delegating tasks without surrendering control, making trusted credential custody the new security frontier.