Concerns are swirling about consumers increasingly handing over access to AI agents. Now, security services are trying to catch up.
Password manager 1Password on Thursday released a new tool that lets agents log into personal accounts on your behalf without requiring them to access your credentials.
A new platform called 1Password for Agents accomplishes this by creating a communication system between agents, users, and password managers. When an agent attempts to log in on your behalf, a unique digital key is sent to 1Password to verify the agent’s digital ID. The user will then be alerted about the login attempt and can personally review the request. 1Password then signs the agent in and enters their credentials on behalf of the digital assistant. This allows your digital assistant to log in and complete the task you are performing.
1Password for Agents allows users to choose their preferred agents, such as Anthropic’s Claude Duet, OpenAI’s Dots or ChatGPT Work, Grok Bot, Meta’s Muse, or agents from AI assistant startup Town. The company says it is working with other companies to expand its services.
“Giving up passwords, secrets, and credentials is a common shortcut to agents, but a copy is saved for next time, potentially leaving sensitive information in a place you have no control over,” the company said in a blog post. “1Password for Agents gives people a reliable way to delegate tasks to personal AI agents while managing the passwords, secrets, and credentials needed for those tasks.”
Online agent activity has made major institutions increasingly unstable. In fact, 1Password’s new platform launches on the same day that Axios reported that U.S. banks began detecting agent activity within user accounts.
This is not necessarily a case of a rogue agent stealing an account, but rather an example of how consumers are increasingly allowing automated services to act on their behalf, including with financial matters. However, the potential for agent-related issues is still large. The big problem seems to be that most banks cannot determine whether an agent’s activity is associated with a user’s request.
Bankers interviewed by the magazine called the situation “horrifying,” and a Chase spokesperson interviewed by Axios also told the magazine that customers “should not give out their usernames and passwords to anyone.”
If you make a purchase through links in our articles, we may earn a small commission. This does not affect editorial independence.
