1Password released a new system on Thursday that lets AI agents log into user accounts without ever seeing the passwords.
The tool, called 1Password for Agents, sits between the digital assistant and the user. When an agent tries to access an account, the password manager sends a unique digital key to verify the agent’s identity. The user then receives a notification and must personally approve the request. Once confirmed, 1Password logs the agent in and fills in the necessary details, allowing the task to finish while the credentials remain hidden.
Supported agents
Users can connect services from Anthropic’s Claude Duet, OpenAI’s Dots or ChatGPT Work, Grok Bot, Meta’s Muse, or agents from Town. The company is currently expanding partnerships with other providers.
In a blog post, the firm noted that handing over credentials is a common shortcut but leaves sensitive data in places the user cannot control once a copy is saved. 1Password for Agents offers a way to delegate work while keeping passwords, secrets, and credentials secure.
Timing matters
The launch coincides with reports that US banks have started detecting agentic activity inside user accounts. Axios says financial institutions are seeing automated services acting on behalf of customers, including on money matters.
This is not necessarily about rogue agents stealing accounts. It is about consumers allowing automated tools to act for them. The difficulty for banks is distinguishing between activity tied to a user’s request and activity that happens without one.
Bankers interviewed by the outlet described the situation as “horrifying,” and a Chase spokesperson told Axios that customers “should not give their usernames and passwords to anyone.”
What it means
The industry is moving toward a model where users approve actions rather than sharing secrets. This shifts the burden of verification from the agent to the person using the service.
