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

Concerns person swirled implicit the entree that consumers are progressively entrusting to AI agents. Now, information services are trying to drawback up.

Password manager 1Password connected Thursday launched a caller instrumentality by which users tin let agents to log into their idiosyncratic accounts connected their behalf, without the cause ever having entree to a user’s credentials.

The caller platform, dubbed 1Password for Agents, accomplishes this by creating a strategy of connection betwixt the agent, the user, and the password manager. When an cause attempts to log successful connected a user’s behalf, a unsocial integer cardinal is sent to 1Password to corroborate the agent’s integer identity. The idiosyncratic is past sent an alert astir the login effort and tin personally corroborate the request. 1Password past signs successful the agent, filling successful the credentials connected the integer assistant’s behalf, which allows it to log successful and implicit immoderate task is astatine hand.

1Password for Agents allows users to take their cause of prime — whether that’s Claude Duet from Anthropic, OpenAI’s Dots oregon ChatGPT Work, Grok Bot, Meta’s Muse, oregon agents from Town, which is an AI adjunct startup. The institution says it is moving with different companies to grow the service.

“Handing implicit a password, secret, oregon credential is simply a communal shortcut to an agent, but tin permission radical with delicate accusation successful places they tin nary longer power arsenic a transcript gets saved for the adjacent time,” the institution said in a blog post. “1Password for Agents gives radical a trusted mode to delegate tasks to idiosyncratic AI agents portion keeping power of the passwords, secrets, and credentials those tasks require.”

Online agentic enactment has progressively unsettled large institutions. Indeed, 1Password’s caller level is launching connected the aforesaid time that Axios reports that U.S. banks person begun to observe agentic enactment wrong idiosyncratic accounts.

This isn’t needfully a lawsuit of rogue agents pilfering accounts but, rather, an illustration of however consumers are progressively allowing automated services to enactment connected their behalf, including connected fiscal matters. However, the imaginable for agent-related problems is inactive large. The large contented appears to beryllium that astir banks cannot archer if the agentic enactment is tied to idiosyncratic requests oregon not.

Bankers interviewed by the outlet described the concern arsenic “horrifying,” and a Chase spokesperson interviewed by Axios additionally told the outlet that customers “should not springiness their usernames and passwords to anyone.”