1Password Launches Tool for Delegating AI Logins

American banks are detecting connections made by agents, without always knowing if they are initiated by customers. In this context, 1Password is launching a tool that allows an agent to log in on behalf of a user, without sharing their credentials.

Banks Describe a "Horrifying" and Hard-to-Define Situation

Bankers interviewed describe the situation as "horrifying." A spokesperson for Chase asserts that customers should not share their usernames and passwords with anyone. Most banks struggle to determine whether agent activity is related to a user's request or not. The potential for issues related to agents remains significant.

1Password Organizes Delegated Authentication with User Approval

1Password launched 1Password for Agents on Thursday, a platform that allows an AI agent to log into a user's account without accessing their credentials. The service establishes a link between the agent, the user, and the password manager. When an agent attempts to access the account, 1Password receives a distinct digital key to verify the digital identity of that agent. A notification is sent to the user, who can then authorize the request. Once approved, 1Password provides the credentials to the agent, who can then complete the task. According to the company, directly handing over a password or secret to an agent can leave sensitive information in places outside the user's control, as a copy can be saved for later use. 1Password presents its tool as a way to delegate tasks to AI agents while maintaining control over the passwords, secrets, and credentials needed.

Compatible Agents and Service Expansion Announced

Users can choose agents such as Claude Duet from Anthropic, Dots, ChatGPT Work from OpenAI, Grok Bot, Muse from Meta, or agents from Town. 1Password indicates that it is working with other companies to expand the service's compatibility.

A Launch at a Time When Agent Activity is Intensifying

The launch of 1Password for Agents comes on the day it is reported that American banks are beginning to detect agent activities on user accounts. Online agent activity is increasingly disrupting major institutions. These activities are not necessarily the work of malicious agents but illustrate how consumers are allowing automated services to act on their behalf, including for financial transactions. Concerns are emerging around this expanded access granted to agents, while security services seek to adapt.