Password manager 1Password has launched 1Password for Agents, a tool that lets a personal AI agent log into user accounts without ever receiving the credentials. The flow relies on a unique digital key for agent identity, a user approval alert, and autofill by the vault itself. The release matters because banks are now detecting agent activity inside accounts without knowing if users authorized it.
How the delegated login works
When an agent attempts to sign in on behalf of a user, it sends a unique digital key to 1Password to confirm its digital identity. The user then receives an alert about the login attempt and can personally confirm the request. Only after that confirmation does 1Password fill in the credentials for the assistant, which then proceeds with the assigned task.
The design removes the common shortcut of pasting a password, secret, or credential directly into an agent. 1Password argues that such copies remain stored for future use in places users no longer control. Under the new scheme, delegation keeps passwords, secrets, and credentials inside the manager while the agent gets session access to complete the work.
At launch, users can choose among several supported assistants, including Claude Duet from Anthropic, OpenAI's Dots and ChatGPT Work, Grok Bot, Meta's Muse, and agents from Town, an AI assistant startup. The company says it is working with other companies to expand coverage. The launch took place on Thursday, the same day Axios reported that U. S. banks had begun detecting agentic activity inside user accounts.
What this means for companies using agents
For businesses experimenting with personal agents for purchases, bookings, or account tasks, the model separates permission from possession of secrets. Security teams no longer need to accept credentials scattered across prompts, scripts, and agent memory to get automation running. Small firms gain a controlled way to trial delegation, while larger organizations get an approval step that fits existing access-review habits.
The limits center on visibility and vendor support. Most banks still cannot distinguish agent actions requested by a customer from unauthorized automation, a gap bankers described as horrifying in the Axios reporting. A Chase spokesperson said customers should not give usernames and passwords to anyone. Buyers should therefore verify which agents and sites are supported, how approvals are logged, and what the news alone does not prove about fraud liability.
A practical marker to watch is whether banks and other account providers start recognizing 1Password-signed agent logins as user-authorized sessions. If detection shifts from unexplained agentic activity to labeled, approved delegation, finance and support teams will face fewer blocks and disputes. If not, delegated login will remain a cleaner credential habit without resolving the identification problem.
