1Password launches a way for AI agents to log in on your behalf without seeing your password

1Password has introduced a new tool called 1Password for Agents that enables AI agents to log into users' personal accounts without gaining access to their passwords. This new platform works by establishing a communication system between the user, the AI agent, and the password manager. When an agent tries to log in on behalf of a user, 1Password for Agents creates a unique digital key to verify the agent's identity.

The user receives an alert about the login attempt and can confirm it. Once verified, the password manager logs in the agent and fills in the necessary credentials on its behalf, allowing the AI to complete the task. Users can choose their preferred agent, such as Claude Duet from Anthropic, OpenAI's Dots, or ChatGPT Work, Grok Bot, Meta's Muse, or agents from Town, an AI assistant startup.

The company is working with other companies to expand the service. 1Password warns that handing over passwords or credentials can be risky, as it leaves sensitive information in places that cannot be controlled. The company's new tool offers a safer way to delegate tasks to AI agents while maintaining control over the passwords and credentials needed for those tasks.

The launch of 1Password for Agents comes as major institutions grapple with the rise of agentic activity online. U.S. banks have reported detecting agentic activity inside user accounts, which could be from rogue agents or consumers allowing automated services to act on their behalf. However, banks struggle to distinguish between user-initiated and agent-initiated activity, calling the situation "horrifying." Some banks advise customers against sharing their usernames and passwords with anyone.